Model save
Browse files- README.md +77 -0
- all_results.json +9 -0
- generation_config.json +12 -0
- model-00001-of-00004.safetensors +1 -1
- model-00002-of-00004.safetensors +1 -1
- model-00003-of-00004.safetensors +1 -1
- model-00004-of-00004.safetensors +1 -1
- runs/Sep12_07-12-04_jjb_prism_dev2/events.out.tfevents.1726093301.jjb_prism_dev2.69652.0 +2 -2
- train_results.json +9 -0
- trainer_state.json +2157 -0
README.md
ADDED
|
@@ -0,0 +1,77 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
---
|
| 2 |
+
library_name: transformers
|
| 3 |
+
license: llama3.1
|
| 4 |
+
base_model: meta-llama/Meta-Llama-3.1-8B-Instruct
|
| 5 |
+
tags:
|
| 6 |
+
- trl
|
| 7 |
+
- cpo
|
| 8 |
+
- generated_from_trainer
|
| 9 |
+
model-index:
|
| 10 |
+
- name: llama3.1-cpo-full-0911
|
| 11 |
+
results: []
|
| 12 |
+
---
|
| 13 |
+
|
| 14 |
+
<!-- This model card has been generated automatically according to the information the Trainer had access to. You
|
| 15 |
+
should probably proofread and complete it, then remove this comment. -->
|
| 16 |
+
|
| 17 |
+
# llama3.1-cpo-full-0911
|
| 18 |
+
|
| 19 |
+
This model is a fine-tuned version of [meta-llama/Meta-Llama-3.1-8B-Instruct](https://huggingface.co/meta-llama/Meta-Llama-3.1-8B-Instruct) on an unknown dataset.
|
| 20 |
+
It achieves the following results on the evaluation set:
|
| 21 |
+
- Loss: 1.5984
|
| 22 |
+
- Rewards/chosen: -14.3945
|
| 23 |
+
- Rewards/rejected: -15.5836
|
| 24 |
+
- Rewards/accuracies: 0.6304
|
| 25 |
+
- Rewards/margins: 1.1892
|
| 26 |
+
- Logps/rejected: -155.8365
|
| 27 |
+
- Logps/chosen: -143.9448
|
| 28 |
+
- Logits/rejected: -0.3142
|
| 29 |
+
- Logits/chosen: -0.3408
|
| 30 |
+
- Nll Loss: 0.3937
|
| 31 |
+
|
| 32 |
+
## Model description
|
| 33 |
+
|
| 34 |
+
More information needed
|
| 35 |
+
|
| 36 |
+
## Intended uses & limitations
|
| 37 |
+
|
| 38 |
+
More information needed
|
| 39 |
+
|
| 40 |
+
## Training and evaluation data
|
| 41 |
+
|
| 42 |
+
More information needed
|
| 43 |
+
|
| 44 |
+
## Training procedure
|
| 45 |
+
|
| 46 |
+
### Training hyperparameters
|
| 47 |
+
|
| 48 |
+
The following hyperparameters were used during training:
|
| 49 |
+
- learning_rate: 1e-06
|
| 50 |
+
- train_batch_size: 4
|
| 51 |
+
- eval_batch_size: 4
|
| 52 |
+
- seed: 42
|
| 53 |
+
- distributed_type: multi-GPU
|
| 54 |
+
- num_devices: 4
|
| 55 |
+
- gradient_accumulation_steps: 8
|
| 56 |
+
- total_train_batch_size: 128
|
| 57 |
+
- total_eval_batch_size: 16
|
| 58 |
+
- optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
|
| 59 |
+
- lr_scheduler_type: linear
|
| 60 |
+
- lr_scheduler_warmup_ratio: 0.1
|
| 61 |
+
- num_epochs: 3
|
| 62 |
+
|
| 63 |
+
### Training results
|
| 64 |
+
|
| 65 |
+
| Training Loss | Epoch | Step | Validation Loss | Rewards/chosen | Rewards/rejected | Rewards/accuracies | Rewards/margins | Logps/rejected | Logps/chosen | Logits/rejected | Logits/chosen | Nll Loss |
|
| 66 |
+
|:-------------:|:------:|:----:|:---------------:|:--------------:|:----------------:|:------------------:|:---------------:|:--------------:|:------------:|:---------------:|:-------------:|:--------:|
|
| 67 |
+
| 1.5867 | 0.9986 | 432 | 1.5248 | -16.0094 | -16.9746 | 0.6587 | 0.9652 | -169.7457 | -160.0941 | -0.4783 | -0.5128 | 0.4373 |
|
| 68 |
+
| 0.7108 | 1.9994 | 865 | 1.5252 | -14.8375 | -15.9459 | 0.6500 | 1.1084 | -159.4588 | -148.3749 | -0.4403 | -0.4684 | 0.4056 |
|
| 69 |
+
| 0.4426 | 2.9957 | 1296 | 1.5984 | -14.3945 | -15.5836 | 0.6304 | 1.1892 | -155.8365 | -143.9448 | -0.3142 | -0.3408 | 0.3937 |
|
| 70 |
+
|
| 71 |
+
|
| 72 |
+
### Framework versions
|
| 73 |
+
|
| 74 |
+
- Transformers 4.44.2
|
| 75 |
+
- Pytorch 2.3.1
|
| 76 |
+
- Datasets 2.21.0
|
| 77 |
+
- Tokenizers 0.19.1
|
all_results.json
ADDED
|
@@ -0,0 +1,9 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"epoch": 2.9956659924877203,
|
| 3 |
+
"total_flos": 0.0,
|
| 4 |
+
"train_loss": 1.025652496903031,
|
| 5 |
+
"train_runtime": 27913.7357,
|
| 6 |
+
"train_samples": 55376,
|
| 7 |
+
"train_samples_per_second": 5.951,
|
| 8 |
+
"train_steps_per_second": 0.046
|
| 9 |
+
}
|
generation_config.json
ADDED
|
@@ -0,0 +1,12 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"bos_token_id": 128000,
|
| 3 |
+
"do_sample": true,
|
| 4 |
+
"eos_token_id": [
|
| 5 |
+
128001,
|
| 6 |
+
128008,
|
| 7 |
+
128009
|
| 8 |
+
],
|
| 9 |
+
"temperature": 0.6,
|
| 10 |
+
"top_p": 0.9,
|
| 11 |
+
"transformers_version": "4.44.2"
|
| 12 |
+
}
|
model-00001-of-00004.safetensors
CHANGED
|
@@ -1,3 +1,3 @@
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
-
oid sha256:
|
| 3 |
size 4976698672
|
|
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:1cc17bf32f2fa6e5e833522d444013103b7c7ba68c9b9e9ce0756f8b98c561c5
|
| 3 |
size 4976698672
|
model-00002-of-00004.safetensors
CHANGED
|
@@ -1,3 +1,3 @@
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
-
oid sha256:
|
| 3 |
size 4999802720
|
|
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:63c2cddbb6ed41d54eda5ebe6d4b7afa3cf04c1b67c2b6924294fbe58457b7f4
|
| 3 |
size 4999802720
|
model-00003-of-00004.safetensors
CHANGED
|
@@ -1,3 +1,3 @@
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
-
oid sha256:
|
| 3 |
size 4915916176
|
|
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:5a4a57d43f6b041b56d7f1628457426a107b7c642c0c48c0d62ca8f18c639fd4
|
| 3 |
size 4915916176
|
model-00004-of-00004.safetensors
CHANGED
|
@@ -1,3 +1,3 @@
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
-
oid sha256:
|
| 3 |
size 1168138808
|
|
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:cba5cff84e8fee0ad79657c241f60a87d8576cef1e95da1a98d86eca6eb28de3
|
| 3 |
size 1168138808
|
runs/Sep12_07-12-04_jjb_prism_dev2/events.out.tfevents.1726093301.jjb_prism_dev2.69652.0
CHANGED
|
@@ -1,3 +1,3 @@
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
-
oid sha256:
|
| 3 |
-
size
|
|
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:a605afdc3eab3af17060e499bcb855b0bed46e26b8960dd121930974645a7e83
|
| 3 |
+
size 103879
|
train_results.json
ADDED
|
@@ -0,0 +1,9 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"epoch": 2.9956659924877203,
|
| 3 |
+
"total_flos": 0.0,
|
| 4 |
+
"train_loss": 1.025652496903031,
|
| 5 |
+
"train_runtime": 27913.7357,
|
| 6 |
+
"train_samples": 55376,
|
| 7 |
+
"train_samples_per_second": 5.951,
|
| 8 |
+
"train_steps_per_second": 0.046
|
| 9 |
+
}
|
trainer_state.json
ADDED
|
@@ -0,0 +1,2157 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"best_metric": null,
|
| 3 |
+
"best_model_checkpoint": null,
|
| 4 |
+
"epoch": 2.9956659924877203,
|
| 5 |
+
"eval_steps": 500,
|
| 6 |
+
"global_step": 1296,
|
| 7 |
+
"is_hyper_param_search": false,
|
| 8 |
+
"is_local_process_zero": true,
|
| 9 |
+
"is_world_process_zero": true,
|
| 10 |
+
"log_history": [
|
| 11 |
+
{
|
| 12 |
+
"epoch": 0.023114706732158336,
|
| 13 |
+
"grad_norm": 66.88623809814453,
|
| 14 |
+
"learning_rate": 7.692307692307692e-08,
|
| 15 |
+
"logits/chosen": -0.3347979187965393,
|
| 16 |
+
"logits/rejected": -0.31468525528907776,
|
| 17 |
+
"logps/chosen": -269.48956298828125,
|
| 18 |
+
"logps/rejected": -267.7704162597656,
|
| 19 |
+
"loss": 2.6159,
|
| 20 |
+
"nll_loss": 0.7417957186698914,
|
| 21 |
+
"rewards/accuracies": 0.4937500059604645,
|
| 22 |
+
"rewards/chosen": -26.94895362854004,
|
| 23 |
+
"rewards/margins": -0.1719130575656891,
|
| 24 |
+
"rewards/rejected": -26.777042388916016,
|
| 25 |
+
"step": 10
|
| 26 |
+
},
|
| 27 |
+
{
|
| 28 |
+
"epoch": 0.04622941346431667,
|
| 29 |
+
"grad_norm": 57.376678466796875,
|
| 30 |
+
"learning_rate": 1.5384615384615385e-07,
|
| 31 |
+
"logits/chosen": -0.3397385776042938,
|
| 32 |
+
"logits/rejected": -0.3220021724700928,
|
| 33 |
+
"logps/chosen": -263.6553039550781,
|
| 34 |
+
"logps/rejected": -270.1563415527344,
|
| 35 |
+
"loss": 2.5433,
|
| 36 |
+
"nll_loss": 0.7268518805503845,
|
| 37 |
+
"rewards/accuracies": 0.5687500238418579,
|
| 38 |
+
"rewards/chosen": -26.365528106689453,
|
| 39 |
+
"rewards/margins": 0.650107741355896,
|
| 40 |
+
"rewards/rejected": -27.015636444091797,
|
| 41 |
+
"step": 20
|
| 42 |
+
},
|
| 43 |
+
{
|
| 44 |
+
"epoch": 0.06934412019647501,
|
| 45 |
+
"grad_norm": 61.90145492553711,
|
| 46 |
+
"learning_rate": 2.3076923076923078e-07,
|
| 47 |
+
"logits/chosen": -0.283037930727005,
|
| 48 |
+
"logits/rejected": -0.2716462016105652,
|
| 49 |
+
"logps/chosen": -260.65875244140625,
|
| 50 |
+
"logps/rejected": -264.00567626953125,
|
| 51 |
+
"loss": 2.4476,
|
| 52 |
+
"nll_loss": 0.7402013540267944,
|
| 53 |
+
"rewards/accuracies": 0.518750011920929,
|
| 54 |
+
"rewards/chosen": -26.065876007080078,
|
| 55 |
+
"rewards/margins": 0.33469510078430176,
|
| 56 |
+
"rewards/rejected": -26.40056800842285,
|
| 57 |
+
"step": 30
|
| 58 |
+
},
|
| 59 |
+
{
|
| 60 |
+
"epoch": 0.09245882692863334,
|
| 61 |
+
"grad_norm": 50.3497428894043,
|
| 62 |
+
"learning_rate": 3.076923076923077e-07,
|
| 63 |
+
"logits/chosen": -0.3415708541870117,
|
| 64 |
+
"logits/rejected": -0.33246317505836487,
|
| 65 |
+
"logps/chosen": -246.55178833007812,
|
| 66 |
+
"logps/rejected": -248.4236297607422,
|
| 67 |
+
"loss": 2.377,
|
| 68 |
+
"nll_loss": 0.748863697052002,
|
| 69 |
+
"rewards/accuracies": 0.5218750238418579,
|
| 70 |
+
"rewards/chosen": -24.655179977416992,
|
| 71 |
+
"rewards/margins": 0.18718396127223969,
|
| 72 |
+
"rewards/rejected": -24.842365264892578,
|
| 73 |
+
"step": 40
|
| 74 |
+
},
|
| 75 |
+
{
|
| 76 |
+
"epoch": 0.11557353366079168,
|
| 77 |
+
"grad_norm": 49.670677185058594,
|
| 78 |
+
"learning_rate": 3.8461538461538463e-07,
|
| 79 |
+
"logits/chosen": -0.408900648355484,
|
| 80 |
+
"logits/rejected": -0.38666287064552307,
|
| 81 |
+
"logps/chosen": -251.69113159179688,
|
| 82 |
+
"logps/rejected": -250.1224822998047,
|
| 83 |
+
"loss": 2.5031,
|
| 84 |
+
"nll_loss": 0.6844735145568848,
|
| 85 |
+
"rewards/accuracies": 0.5,
|
| 86 |
+
"rewards/chosen": -25.169116973876953,
|
| 87 |
+
"rewards/margins": -0.15686817467212677,
|
| 88 |
+
"rewards/rejected": -25.01224708557129,
|
| 89 |
+
"step": 50
|
| 90 |
+
},
|
| 91 |
+
{
|
| 92 |
+
"epoch": 0.13868824039295002,
|
| 93 |
+
"grad_norm": 56.84200668334961,
|
| 94 |
+
"learning_rate": 4.6153846153846156e-07,
|
| 95 |
+
"logits/chosen": -0.5338214635848999,
|
| 96 |
+
"logits/rejected": -0.5284141302108765,
|
| 97 |
+
"logps/chosen": -228.7010498046875,
|
| 98 |
+
"logps/rejected": -227.7324676513672,
|
| 99 |
+
"loss": 2.3944,
|
| 100 |
+
"nll_loss": 0.6640177965164185,
|
| 101 |
+
"rewards/accuracies": 0.5249999761581421,
|
| 102 |
+
"rewards/chosen": -22.870105743408203,
|
| 103 |
+
"rewards/margins": -0.09685804694890976,
|
| 104 |
+
"rewards/rejected": -22.77324867248535,
|
| 105 |
+
"step": 60
|
| 106 |
+
},
|
| 107 |
+
{
|
| 108 |
+
"epoch": 0.16180294712510834,
|
| 109 |
+
"grad_norm": 54.86004638671875,
|
| 110 |
+
"learning_rate": 5.384615384615384e-07,
|
| 111 |
+
"logits/chosen": -0.7272528409957886,
|
| 112 |
+
"logits/rejected": -0.7068361639976501,
|
| 113 |
+
"logps/chosen": -207.84902954101562,
|
| 114 |
+
"logps/rejected": -207.88827514648438,
|
| 115 |
+
"loss": 2.3592,
|
| 116 |
+
"nll_loss": 0.6152782440185547,
|
| 117 |
+
"rewards/accuracies": 0.4937500059604645,
|
| 118 |
+
"rewards/chosen": -20.78490447998047,
|
| 119 |
+
"rewards/margins": 0.0039252908900380135,
|
| 120 |
+
"rewards/rejected": -20.788829803466797,
|
| 121 |
+
"step": 70
|
| 122 |
+
},
|
| 123 |
+
{
|
| 124 |
+
"epoch": 0.1849176538572667,
|
| 125 |
+
"grad_norm": 51.230716705322266,
|
| 126 |
+
"learning_rate": 6.153846153846154e-07,
|
| 127 |
+
"logits/chosen": -0.8364453315734863,
|
| 128 |
+
"logits/rejected": -0.8120291829109192,
|
| 129 |
+
"logps/chosen": -197.1513671875,
|
| 130 |
+
"logps/rejected": -198.78500366210938,
|
| 131 |
+
"loss": 2.1038,
|
| 132 |
+
"nll_loss": 0.5486661195755005,
|
| 133 |
+
"rewards/accuracies": 0.528124988079071,
|
| 134 |
+
"rewards/chosen": -19.715137481689453,
|
| 135 |
+
"rewards/margins": 0.16336408257484436,
|
| 136 |
+
"rewards/rejected": -19.87849998474121,
|
| 137 |
+
"step": 80
|
| 138 |
+
},
|
| 139 |
+
{
|
| 140 |
+
"epoch": 0.208032360589425,
|
| 141 |
+
"grad_norm": 49.48822784423828,
|
| 142 |
+
"learning_rate": 6.923076923076922e-07,
|
| 143 |
+
"logits/chosen": -0.6982623338699341,
|
| 144 |
+
"logits/rejected": -0.6783713698387146,
|
| 145 |
+
"logps/chosen": -170.0928497314453,
|
| 146 |
+
"logps/rejected": -176.90493774414062,
|
| 147 |
+
"loss": 1.9408,
|
| 148 |
+
"nll_loss": 0.47356802225112915,
|
| 149 |
+
"rewards/accuracies": 0.5874999761581421,
|
| 150 |
+
"rewards/chosen": -17.00928497314453,
|
| 151 |
+
"rewards/margins": 0.6812085509300232,
|
| 152 |
+
"rewards/rejected": -17.690494537353516,
|
| 153 |
+
"step": 90
|
| 154 |
+
},
|
| 155 |
+
{
|
| 156 |
+
"epoch": 0.23114706732158335,
|
| 157 |
+
"grad_norm": 51.29805374145508,
|
| 158 |
+
"learning_rate": 7.692307692307693e-07,
|
| 159 |
+
"logits/chosen": -0.5311590433120728,
|
| 160 |
+
"logits/rejected": -0.5080990195274353,
|
| 161 |
+
"logps/chosen": -162.14718627929688,
|
| 162 |
+
"logps/rejected": -163.9295196533203,
|
| 163 |
+
"loss": 2.045,
|
| 164 |
+
"nll_loss": 0.44545722007751465,
|
| 165 |
+
"rewards/accuracies": 0.5249999761581421,
|
| 166 |
+
"rewards/chosen": -16.214717864990234,
|
| 167 |
+
"rewards/margins": 0.17823390662670135,
|
| 168 |
+
"rewards/rejected": -16.3929500579834,
|
| 169 |
+
"step": 100
|
| 170 |
+
},
|
| 171 |
+
{
|
| 172 |
+
"epoch": 0.2542617740537417,
|
| 173 |
+
"grad_norm": 48.0262565612793,
|
| 174 |
+
"learning_rate": 8.461538461538461e-07,
|
| 175 |
+
"logits/chosen": -0.40089306235313416,
|
| 176 |
+
"logits/rejected": -0.37224632501602173,
|
| 177 |
+
"logps/chosen": -149.81529235839844,
|
| 178 |
+
"logps/rejected": -153.31297302246094,
|
| 179 |
+
"loss": 1.8155,
|
| 180 |
+
"nll_loss": 0.41241997480392456,
|
| 181 |
+
"rewards/accuracies": 0.512499988079071,
|
| 182 |
+
"rewards/chosen": -14.981531143188477,
|
| 183 |
+
"rewards/margins": 0.3497660458087921,
|
| 184 |
+
"rewards/rejected": -15.331296920776367,
|
| 185 |
+
"step": 110
|
| 186 |
+
},
|
| 187 |
+
{
|
| 188 |
+
"epoch": 0.27737648078590005,
|
| 189 |
+
"grad_norm": 47.4226188659668,
|
| 190 |
+
"learning_rate": 9.230769230769231e-07,
|
| 191 |
+
"logits/chosen": -0.3508472144603729,
|
| 192 |
+
"logits/rejected": -0.3243268132209778,
|
| 193 |
+
"logps/chosen": -155.70640563964844,
|
| 194 |
+
"logps/rejected": -155.49754333496094,
|
| 195 |
+
"loss": 1.8406,
|
| 196 |
+
"nll_loss": 0.42402368783950806,
|
| 197 |
+
"rewards/accuracies": 0.5218750238418579,
|
| 198 |
+
"rewards/chosen": -15.570638656616211,
|
| 199 |
+
"rewards/margins": -0.020885681733489037,
|
| 200 |
+
"rewards/rejected": -15.549756050109863,
|
| 201 |
+
"step": 120
|
| 202 |
+
},
|
| 203 |
+
{
|
| 204 |
+
"epoch": 0.30049118751805837,
|
| 205 |
+
"grad_norm": 55.037635803222656,
|
| 206 |
+
"learning_rate": 1e-06,
|
| 207 |
+
"logits/chosen": -0.39025238156318665,
|
| 208 |
+
"logits/rejected": -0.36617571115493774,
|
| 209 |
+
"logps/chosen": -152.71519470214844,
|
| 210 |
+
"logps/rejected": -160.30941772460938,
|
| 211 |
+
"loss": 1.8955,
|
| 212 |
+
"nll_loss": 0.42156749963760376,
|
| 213 |
+
"rewards/accuracies": 0.5718749761581421,
|
| 214 |
+
"rewards/chosen": -15.271520614624023,
|
| 215 |
+
"rewards/margins": 0.7594239711761475,
|
| 216 |
+
"rewards/rejected": -16.030942916870117,
|
| 217 |
+
"step": 130
|
| 218 |
+
},
|
| 219 |
+
{
|
| 220 |
+
"epoch": 0.3236058942502167,
|
| 221 |
+
"grad_norm": 59.5285530090332,
|
| 222 |
+
"learning_rate": 9.914236706689536e-07,
|
| 223 |
+
"logits/chosen": -0.4246004521846771,
|
| 224 |
+
"logits/rejected": -0.4167487621307373,
|
| 225 |
+
"logps/chosen": -142.45510864257812,
|
| 226 |
+
"logps/rejected": -147.0147247314453,
|
| 227 |
+
"loss": 1.8864,
|
| 228 |
+
"nll_loss": 0.40775737166404724,
|
| 229 |
+
"rewards/accuracies": 0.574999988079071,
|
| 230 |
+
"rewards/chosen": -14.245511054992676,
|
| 231 |
+
"rewards/margins": 0.4559602737426758,
|
| 232 |
+
"rewards/rejected": -14.701472282409668,
|
| 233 |
+
"step": 140
|
| 234 |
+
},
|
| 235 |
+
{
|
| 236 |
+
"epoch": 0.34672060098237506,
|
| 237 |
+
"grad_norm": 65.72272491455078,
|
| 238 |
+
"learning_rate": 9.828473413379073e-07,
|
| 239 |
+
"logits/chosen": -0.41359296441078186,
|
| 240 |
+
"logits/rejected": -0.4051112234592438,
|
| 241 |
+
"logps/chosen": -144.20643615722656,
|
| 242 |
+
"logps/rejected": -148.5088653564453,
|
| 243 |
+
"loss": 1.9229,
|
| 244 |
+
"nll_loss": 0.41948890686035156,
|
| 245 |
+
"rewards/accuracies": 0.578125,
|
| 246 |
+
"rewards/chosen": -14.420641899108887,
|
| 247 |
+
"rewards/margins": 0.43024420738220215,
|
| 248 |
+
"rewards/rejected": -14.850886344909668,
|
| 249 |
+
"step": 150
|
| 250 |
+
},
|
| 251 |
+
{
|
| 252 |
+
"epoch": 0.3698353077145334,
|
| 253 |
+
"grad_norm": 44.7535400390625,
|
| 254 |
+
"learning_rate": 9.74271012006861e-07,
|
| 255 |
+
"logits/chosen": -0.4151730537414551,
|
| 256 |
+
"logits/rejected": -0.3891311287879944,
|
| 257 |
+
"logps/chosen": -159.87423706054688,
|
| 258 |
+
"logps/rejected": -162.4603729248047,
|
| 259 |
+
"loss": 1.7633,
|
| 260 |
+
"nll_loss": 0.4239375591278076,
|
| 261 |
+
"rewards/accuracies": 0.543749988079071,
|
| 262 |
+
"rewards/chosen": -15.987421989440918,
|
| 263 |
+
"rewards/margins": 0.25861597061157227,
|
| 264 |
+
"rewards/rejected": -16.246036529541016,
|
| 265 |
+
"step": 160
|
| 266 |
+
},
|
| 267 |
+
{
|
| 268 |
+
"epoch": 0.3929500144466917,
|
| 269 |
+
"grad_norm": 54.54914093017578,
|
| 270 |
+
"learning_rate": 9.656946826758147e-07,
|
| 271 |
+
"logits/chosen": -0.4074994921684265,
|
| 272 |
+
"logits/rejected": -0.39586925506591797,
|
| 273 |
+
"logps/chosen": -152.30300903320312,
|
| 274 |
+
"logps/rejected": -158.6874237060547,
|
| 275 |
+
"loss": 1.6418,
|
| 276 |
+
"nll_loss": 0.4169434607028961,
|
| 277 |
+
"rewards/accuracies": 0.574999988079071,
|
| 278 |
+
"rewards/chosen": -15.230302810668945,
|
| 279 |
+
"rewards/margins": 0.6384423971176147,
|
| 280 |
+
"rewards/rejected": -15.868745803833008,
|
| 281 |
+
"step": 170
|
| 282 |
+
},
|
| 283 |
+
{
|
| 284 |
+
"epoch": 0.41606472117885,
|
| 285 |
+
"grad_norm": 50.04910659790039,
|
| 286 |
+
"learning_rate": 9.571183533447683e-07,
|
| 287 |
+
"logits/chosen": -0.35864660143852234,
|
| 288 |
+
"logits/rejected": -0.3550480604171753,
|
| 289 |
+
"logps/chosen": -148.34335327148438,
|
| 290 |
+
"logps/rejected": -154.14390563964844,
|
| 291 |
+
"loss": 1.6776,
|
| 292 |
+
"nll_loss": 0.4253178536891937,
|
| 293 |
+
"rewards/accuracies": 0.565625011920929,
|
| 294 |
+
"rewards/chosen": -14.83433723449707,
|
| 295 |
+
"rewards/margins": 0.5800559520721436,
|
| 296 |
+
"rewards/rejected": -15.414392471313477,
|
| 297 |
+
"step": 180
|
| 298 |
+
},
|
| 299 |
+
{
|
| 300 |
+
"epoch": 0.4391794279110084,
|
| 301 |
+
"grad_norm": 50.81975555419922,
|
| 302 |
+
"learning_rate": 9.485420240137221e-07,
|
| 303 |
+
"logits/chosen": -0.369549036026001,
|
| 304 |
+
"logits/rejected": -0.35568320751190186,
|
| 305 |
+
"logps/chosen": -158.10638427734375,
|
| 306 |
+
"logps/rejected": -163.52792358398438,
|
| 307 |
+
"loss": 1.7469,
|
| 308 |
+
"nll_loss": 0.43363142013549805,
|
| 309 |
+
"rewards/accuracies": 0.5874999761581421,
|
| 310 |
+
"rewards/chosen": -15.810638427734375,
|
| 311 |
+
"rewards/margins": 0.5421568155288696,
|
| 312 |
+
"rewards/rejected": -16.352794647216797,
|
| 313 |
+
"step": 190
|
| 314 |
+
},
|
| 315 |
+
{
|
| 316 |
+
"epoch": 0.4622941346431667,
|
| 317 |
+
"grad_norm": 45.42930221557617,
|
| 318 |
+
"learning_rate": 9.399656946826757e-07,
|
| 319 |
+
"logits/chosen": -0.3367421329021454,
|
| 320 |
+
"logits/rejected": -0.3279821276664734,
|
| 321 |
+
"logps/chosen": -152.1122589111328,
|
| 322 |
+
"logps/rejected": -158.53919982910156,
|
| 323 |
+
"loss": 1.7332,
|
| 324 |
+
"nll_loss": 0.4192579686641693,
|
| 325 |
+
"rewards/accuracies": 0.596875011920929,
|
| 326 |
+
"rewards/chosen": -15.211225509643555,
|
| 327 |
+
"rewards/margins": 0.6426929235458374,
|
| 328 |
+
"rewards/rejected": -15.853918075561523,
|
| 329 |
+
"step": 200
|
| 330 |
+
},
|
| 331 |
+
{
|
| 332 |
+
"epoch": 0.48540884137532503,
|
| 333 |
+
"grad_norm": 56.53390884399414,
|
| 334 |
+
"learning_rate": 9.313893653516295e-07,
|
| 335 |
+
"logits/chosen": -0.38974902033805847,
|
| 336 |
+
"logits/rejected": -0.3610435128211975,
|
| 337 |
+
"logps/chosen": -150.60289001464844,
|
| 338 |
+
"logps/rejected": -152.52676391601562,
|
| 339 |
+
"loss": 1.68,
|
| 340 |
+
"nll_loss": 0.4199536442756653,
|
| 341 |
+
"rewards/accuracies": 0.5375000238418579,
|
| 342 |
+
"rewards/chosen": -15.06028938293457,
|
| 343 |
+
"rewards/margins": 0.19238761067390442,
|
| 344 |
+
"rewards/rejected": -15.252676010131836,
|
| 345 |
+
"step": 210
|
| 346 |
+
},
|
| 347 |
+
{
|
| 348 |
+
"epoch": 0.5085235481074833,
|
| 349 |
+
"grad_norm": 51.94132995605469,
|
| 350 |
+
"learning_rate": 9.228130360205832e-07,
|
| 351 |
+
"logits/chosen": -0.4055999219417572,
|
| 352 |
+
"logits/rejected": -0.3751467168331146,
|
| 353 |
+
"logps/chosen": -166.01934814453125,
|
| 354 |
+
"logps/rejected": -171.43978881835938,
|
| 355 |
+
"loss": 1.7614,
|
| 356 |
+
"nll_loss": 0.44684529304504395,
|
| 357 |
+
"rewards/accuracies": 0.6312500238418579,
|
| 358 |
+
"rewards/chosen": -16.6019344329834,
|
| 359 |
+
"rewards/margins": 0.5420453548431396,
|
| 360 |
+
"rewards/rejected": -17.143980026245117,
|
| 361 |
+
"step": 220
|
| 362 |
+
},
|
| 363 |
+
{
|
| 364 |
+
"epoch": 0.5316382548396418,
|
| 365 |
+
"grad_norm": 47.48794174194336,
|
| 366 |
+
"learning_rate": 9.142367066895368e-07,
|
| 367 |
+
"logits/chosen": -0.4172639846801758,
|
| 368 |
+
"logits/rejected": -0.39604124426841736,
|
| 369 |
+
"logps/chosen": -161.71559143066406,
|
| 370 |
+
"logps/rejected": -167.9268341064453,
|
| 371 |
+
"loss": 1.7662,
|
| 372 |
+
"nll_loss": 0.43918871879577637,
|
| 373 |
+
"rewards/accuracies": 0.5687500238418579,
|
| 374 |
+
"rewards/chosen": -16.171558380126953,
|
| 375 |
+
"rewards/margins": 0.6211244463920593,
|
| 376 |
+
"rewards/rejected": -16.792682647705078,
|
| 377 |
+
"step": 230
|
| 378 |
+
},
|
| 379 |
+
{
|
| 380 |
+
"epoch": 0.5547529615718001,
|
| 381 |
+
"grad_norm": 47.16508865356445,
|
| 382 |
+
"learning_rate": 9.056603773584905e-07,
|
| 383 |
+
"logits/chosen": -0.4093747138977051,
|
| 384 |
+
"logits/rejected": -0.39282265305519104,
|
| 385 |
+
"logps/chosen": -160.66146850585938,
|
| 386 |
+
"logps/rejected": -166.21859741210938,
|
| 387 |
+
"loss": 1.5991,
|
| 388 |
+
"nll_loss": 0.4364839196205139,
|
| 389 |
+
"rewards/accuracies": 0.5874999761581421,
|
| 390 |
+
"rewards/chosen": -16.06614875793457,
|
| 391 |
+
"rewards/margins": 0.5557123422622681,
|
| 392 |
+
"rewards/rejected": -16.62186050415039,
|
| 393 |
+
"step": 240
|
| 394 |
+
},
|
| 395 |
+
{
|
| 396 |
+
"epoch": 0.5778676683039584,
|
| 397 |
+
"grad_norm": 55.471988677978516,
|
| 398 |
+
"learning_rate": 8.970840480274442e-07,
|
| 399 |
+
"logits/chosen": -0.3893429636955261,
|
| 400 |
+
"logits/rejected": -0.371797651052475,
|
| 401 |
+
"logps/chosen": -157.74957275390625,
|
| 402 |
+
"logps/rejected": -160.80726623535156,
|
| 403 |
+
"loss": 1.7402,
|
| 404 |
+
"nll_loss": 0.41964635252952576,
|
| 405 |
+
"rewards/accuracies": 0.53125,
|
| 406 |
+
"rewards/chosen": -15.774957656860352,
|
| 407 |
+
"rewards/margins": 0.3057698607444763,
|
| 408 |
+
"rewards/rejected": -16.08072853088379,
|
| 409 |
+
"step": 250
|
| 410 |
+
},
|
| 411 |
+
{
|
| 412 |
+
"epoch": 0.6009823750361167,
|
| 413 |
+
"grad_norm": 51.49521255493164,
|
| 414 |
+
"learning_rate": 8.88507718696398e-07,
|
| 415 |
+
"logits/chosen": -0.508994460105896,
|
| 416 |
+
"logits/rejected": -0.49335479736328125,
|
| 417 |
+
"logps/chosen": -152.62657165527344,
|
| 418 |
+
"logps/rejected": -161.82626342773438,
|
| 419 |
+
"loss": 1.5932,
|
| 420 |
+
"nll_loss": 0.4252961277961731,
|
| 421 |
+
"rewards/accuracies": 0.6499999761581421,
|
| 422 |
+
"rewards/chosen": -15.262657165527344,
|
| 423 |
+
"rewards/margins": 0.919969379901886,
|
| 424 |
+
"rewards/rejected": -16.182626724243164,
|
| 425 |
+
"step": 260
|
| 426 |
+
},
|
| 427 |
+
{
|
| 428 |
+
"epoch": 0.624097081768275,
|
| 429 |
+
"grad_norm": 42.050086975097656,
|
| 430 |
+
"learning_rate": 8.799313893653516e-07,
|
| 431 |
+
"logits/chosen": -0.5146248936653137,
|
| 432 |
+
"logits/rejected": -0.5052975416183472,
|
| 433 |
+
"logps/chosen": -153.78634643554688,
|
| 434 |
+
"logps/rejected": -161.01736450195312,
|
| 435 |
+
"loss": 1.6238,
|
| 436 |
+
"nll_loss": 0.4203058183193207,
|
| 437 |
+
"rewards/accuracies": 0.6187499761581421,
|
| 438 |
+
"rewards/chosen": -15.378637313842773,
|
| 439 |
+
"rewards/margins": 0.723101019859314,
|
| 440 |
+
"rewards/rejected": -16.10173797607422,
|
| 441 |
+
"step": 270
|
| 442 |
+
},
|
| 443 |
+
{
|
| 444 |
+
"epoch": 0.6472117885004334,
|
| 445 |
+
"grad_norm": 47.4835205078125,
|
| 446 |
+
"learning_rate": 8.713550600343052e-07,
|
| 447 |
+
"logits/chosen": -0.45883655548095703,
|
| 448 |
+
"logits/rejected": -0.4298950135707855,
|
| 449 |
+
"logps/chosen": -156.53761291503906,
|
| 450 |
+
"logps/rejected": -165.87417602539062,
|
| 451 |
+
"loss": 1.611,
|
| 452 |
+
"nll_loss": 0.4291974604129791,
|
| 453 |
+
"rewards/accuracies": 0.6499999761581421,
|
| 454 |
+
"rewards/chosen": -15.65376091003418,
|
| 455 |
+
"rewards/margins": 0.9336563944816589,
|
| 456 |
+
"rewards/rejected": -16.58741569519043,
|
| 457 |
+
"step": 280
|
| 458 |
+
},
|
| 459 |
+
{
|
| 460 |
+
"epoch": 0.6703264952325917,
|
| 461 |
+
"grad_norm": 52.52881622314453,
|
| 462 |
+
"learning_rate": 8.62778730703259e-07,
|
| 463 |
+
"logits/chosen": -0.5712844133377075,
|
| 464 |
+
"logits/rejected": -0.5348969101905823,
|
| 465 |
+
"logps/chosen": -159.44664001464844,
|
| 466 |
+
"logps/rejected": -164.33367919921875,
|
| 467 |
+
"loss": 1.605,
|
| 468 |
+
"nll_loss": 0.41903525590896606,
|
| 469 |
+
"rewards/accuracies": 0.6000000238418579,
|
| 470 |
+
"rewards/chosen": -15.944664001464844,
|
| 471 |
+
"rewards/margins": 0.4887046813964844,
|
| 472 |
+
"rewards/rejected": -16.433368682861328,
|
| 473 |
+
"step": 290
|
| 474 |
+
},
|
| 475 |
+
{
|
| 476 |
+
"epoch": 0.6934412019647501,
|
| 477 |
+
"grad_norm": 40.79169464111328,
|
| 478 |
+
"learning_rate": 8.542024013722127e-07,
|
| 479 |
+
"logits/chosen": -0.4996193051338196,
|
| 480 |
+
"logits/rejected": -0.4892319142818451,
|
| 481 |
+
"logps/chosen": -161.40255737304688,
|
| 482 |
+
"logps/rejected": -170.85952758789062,
|
| 483 |
+
"loss": 1.6188,
|
| 484 |
+
"nll_loss": 0.43107661604881287,
|
| 485 |
+
"rewards/accuracies": 0.612500011920929,
|
| 486 |
+
"rewards/chosen": -16.140254974365234,
|
| 487 |
+
"rewards/margins": 0.9456993341445923,
|
| 488 |
+
"rewards/rejected": -17.085954666137695,
|
| 489 |
+
"step": 300
|
| 490 |
+
},
|
| 491 |
+
{
|
| 492 |
+
"epoch": 0.7165559086969084,
|
| 493 |
+
"grad_norm": 46.80116653442383,
|
| 494 |
+
"learning_rate": 8.456260720411664e-07,
|
| 495 |
+
"logits/chosen": -0.5246766805648804,
|
| 496 |
+
"logits/rejected": -0.5159187316894531,
|
| 497 |
+
"logps/chosen": -166.08078002929688,
|
| 498 |
+
"logps/rejected": -169.9657440185547,
|
| 499 |
+
"loss": 1.6674,
|
| 500 |
+
"nll_loss": 0.4305740296840668,
|
| 501 |
+
"rewards/accuracies": 0.546875,
|
| 502 |
+
"rewards/chosen": -16.60807991027832,
|
| 503 |
+
"rewards/margins": 0.38849523663520813,
|
| 504 |
+
"rewards/rejected": -16.99657440185547,
|
| 505 |
+
"step": 310
|
| 506 |
+
},
|
| 507 |
+
{
|
| 508 |
+
"epoch": 0.7396706154290668,
|
| 509 |
+
"grad_norm": 44.07027053833008,
|
| 510 |
+
"learning_rate": 8.3704974271012e-07,
|
| 511 |
+
"logits/chosen": -0.5243274569511414,
|
| 512 |
+
"logits/rejected": -0.5057517290115356,
|
| 513 |
+
"logps/chosen": -159.46981811523438,
|
| 514 |
+
"logps/rejected": -168.90744018554688,
|
| 515 |
+
"loss": 1.6418,
|
| 516 |
+
"nll_loss": 0.4369947910308838,
|
| 517 |
+
"rewards/accuracies": 0.6156250238418579,
|
| 518 |
+
"rewards/chosen": -15.946983337402344,
|
| 519 |
+
"rewards/margins": 0.9437603950500488,
|
| 520 |
+
"rewards/rejected": -16.890743255615234,
|
| 521 |
+
"step": 320
|
| 522 |
+
},
|
| 523 |
+
{
|
| 524 |
+
"epoch": 0.7627853221612251,
|
| 525 |
+
"grad_norm": 47.953094482421875,
|
| 526 |
+
"learning_rate": 8.284734133790737e-07,
|
| 527 |
+
"logits/chosen": -0.5663309097290039,
|
| 528 |
+
"logits/rejected": -0.5677110552787781,
|
| 529 |
+
"logps/chosen": -157.61669921875,
|
| 530 |
+
"logps/rejected": -165.10401916503906,
|
| 531 |
+
"loss": 1.632,
|
| 532 |
+
"nll_loss": 0.4376348853111267,
|
| 533 |
+
"rewards/accuracies": 0.625,
|
| 534 |
+
"rewards/chosen": -15.761670112609863,
|
| 535 |
+
"rewards/margins": 0.7487320303916931,
|
| 536 |
+
"rewards/rejected": -16.51040267944336,
|
| 537 |
+
"step": 330
|
| 538 |
+
},
|
| 539 |
+
{
|
| 540 |
+
"epoch": 0.7859000288933834,
|
| 541 |
+
"grad_norm": 49.6301383972168,
|
| 542 |
+
"learning_rate": 8.198970840480274e-07,
|
| 543 |
+
"logits/chosen": -0.6203452348709106,
|
| 544 |
+
"logits/rejected": -0.6113607287406921,
|
| 545 |
+
"logps/chosen": -163.147216796875,
|
| 546 |
+
"logps/rejected": -169.24349975585938,
|
| 547 |
+
"loss": 1.5495,
|
| 548 |
+
"nll_loss": 0.44397497177124023,
|
| 549 |
+
"rewards/accuracies": 0.590624988079071,
|
| 550 |
+
"rewards/chosen": -16.314722061157227,
|
| 551 |
+
"rewards/margins": 0.6096271872520447,
|
| 552 |
+
"rewards/rejected": -16.92435073852539,
|
| 553 |
+
"step": 340
|
| 554 |
+
},
|
| 555 |
+
{
|
| 556 |
+
"epoch": 0.8090147356255417,
|
| 557 |
+
"grad_norm": 49.70254135131836,
|
| 558 |
+
"learning_rate": 8.113207547169812e-07,
|
| 559 |
+
"logits/chosen": -0.6143781542778015,
|
| 560 |
+
"logits/rejected": -0.6132633090019226,
|
| 561 |
+
"logps/chosen": -159.93780517578125,
|
| 562 |
+
"logps/rejected": -169.04864501953125,
|
| 563 |
+
"loss": 1.6389,
|
| 564 |
+
"nll_loss": 0.41862422227859497,
|
| 565 |
+
"rewards/accuracies": 0.59375,
|
| 566 |
+
"rewards/chosen": -15.993780136108398,
|
| 567 |
+
"rewards/margins": 0.9110851287841797,
|
| 568 |
+
"rewards/rejected": -16.904865264892578,
|
| 569 |
+
"step": 350
|
| 570 |
+
},
|
| 571 |
+
{
|
| 572 |
+
"epoch": 0.8321294423577,
|
| 573 |
+
"grad_norm": 53.936744689941406,
|
| 574 |
+
"learning_rate": 8.027444253859347e-07,
|
| 575 |
+
"logits/chosen": -0.5789452791213989,
|
| 576 |
+
"logits/rejected": -0.5579988360404968,
|
| 577 |
+
"logps/chosen": -163.99453735351562,
|
| 578 |
+
"logps/rejected": -169.4527130126953,
|
| 579 |
+
"loss": 1.6607,
|
| 580 |
+
"nll_loss": 0.43801170587539673,
|
| 581 |
+
"rewards/accuracies": 0.596875011920929,
|
| 582 |
+
"rewards/chosen": -16.39945411682129,
|
| 583 |
+
"rewards/margins": 0.545818030834198,
|
| 584 |
+
"rewards/rejected": -16.945270538330078,
|
| 585 |
+
"step": 360
|
| 586 |
+
},
|
| 587 |
+
{
|
| 588 |
+
"epoch": 0.8552441490898585,
|
| 589 |
+
"grad_norm": 49.6528434753418,
|
| 590 |
+
"learning_rate": 7.941680960548884e-07,
|
| 591 |
+
"logits/chosen": -0.6104979515075684,
|
| 592 |
+
"logits/rejected": -0.5787642002105713,
|
| 593 |
+
"logps/chosen": -156.12814331054688,
|
| 594 |
+
"logps/rejected": -166.38809204101562,
|
| 595 |
+
"loss": 1.5211,
|
| 596 |
+
"nll_loss": 0.4215327203273773,
|
| 597 |
+
"rewards/accuracies": 0.653124988079071,
|
| 598 |
+
"rewards/chosen": -15.612813949584961,
|
| 599 |
+
"rewards/margins": 1.0259974002838135,
|
| 600 |
+
"rewards/rejected": -16.638811111450195,
|
| 601 |
+
"step": 370
|
| 602 |
+
},
|
| 603 |
+
{
|
| 604 |
+
"epoch": 0.8783588558220168,
|
| 605 |
+
"grad_norm": 56.63078308105469,
|
| 606 |
+
"learning_rate": 7.855917667238422e-07,
|
| 607 |
+
"logits/chosen": -0.5714787840843201,
|
| 608 |
+
"logits/rejected": -0.5493099093437195,
|
| 609 |
+
"logps/chosen": -159.0956573486328,
|
| 610 |
+
"logps/rejected": -167.24899291992188,
|
| 611 |
+
"loss": 1.5505,
|
| 612 |
+
"nll_loss": 0.44679298996925354,
|
| 613 |
+
"rewards/accuracies": 0.628125011920929,
|
| 614 |
+
"rewards/chosen": -15.909564018249512,
|
| 615 |
+
"rewards/margins": 0.8153331875801086,
|
| 616 |
+
"rewards/rejected": -16.724897384643555,
|
| 617 |
+
"step": 380
|
| 618 |
+
},
|
| 619 |
+
{
|
| 620 |
+
"epoch": 0.9014735625541751,
|
| 621 |
+
"grad_norm": 48.641788482666016,
|
| 622 |
+
"learning_rate": 7.770154373927959e-07,
|
| 623 |
+
"logits/chosen": -0.5352247357368469,
|
| 624 |
+
"logits/rejected": -0.5140178799629211,
|
| 625 |
+
"logps/chosen": -159.83766174316406,
|
| 626 |
+
"logps/rejected": -170.78448486328125,
|
| 627 |
+
"loss": 1.5767,
|
| 628 |
+
"nll_loss": 0.4272715449333191,
|
| 629 |
+
"rewards/accuracies": 0.6468750238418579,
|
| 630 |
+
"rewards/chosen": -15.9837646484375,
|
| 631 |
+
"rewards/margins": 1.0946825742721558,
|
| 632 |
+
"rewards/rejected": -17.078449249267578,
|
| 633 |
+
"step": 390
|
| 634 |
+
},
|
| 635 |
+
{
|
| 636 |
+
"epoch": 0.9245882692863334,
|
| 637 |
+
"grad_norm": 48.1425666809082,
|
| 638 |
+
"learning_rate": 7.684391080617495e-07,
|
| 639 |
+
"logits/chosen": -0.4901934564113617,
|
| 640 |
+
"logits/rejected": -0.46460065245628357,
|
| 641 |
+
"logps/chosen": -157.80355834960938,
|
| 642 |
+
"logps/rejected": -165.04429626464844,
|
| 643 |
+
"loss": 1.5406,
|
| 644 |
+
"nll_loss": 0.4312973618507385,
|
| 645 |
+
"rewards/accuracies": 0.621874988079071,
|
| 646 |
+
"rewards/chosen": -15.780357360839844,
|
| 647 |
+
"rewards/margins": 0.7240732908248901,
|
| 648 |
+
"rewards/rejected": -16.504430770874023,
|
| 649 |
+
"step": 400
|
| 650 |
+
},
|
| 651 |
+
{
|
| 652 |
+
"epoch": 0.9477029760184917,
|
| 653 |
+
"grad_norm": 49.3435173034668,
|
| 654 |
+
"learning_rate": 7.598627787307032e-07,
|
| 655 |
+
"logits/chosen": -0.46225637197494507,
|
| 656 |
+
"logits/rejected": -0.45260077714920044,
|
| 657 |
+
"logps/chosen": -149.41693115234375,
|
| 658 |
+
"logps/rejected": -160.54165649414062,
|
| 659 |
+
"loss": 1.5048,
|
| 660 |
+
"nll_loss": 0.4266100823879242,
|
| 661 |
+
"rewards/accuracies": 0.625,
|
| 662 |
+
"rewards/chosen": -14.941696166992188,
|
| 663 |
+
"rewards/margins": 1.1124706268310547,
|
| 664 |
+
"rewards/rejected": -16.05416488647461,
|
| 665 |
+
"step": 410
|
| 666 |
+
},
|
| 667 |
+
{
|
| 668 |
+
"epoch": 0.9708176827506501,
|
| 669 |
+
"grad_norm": 44.590450286865234,
|
| 670 |
+
"learning_rate": 7.512864493996569e-07,
|
| 671 |
+
"logits/chosen": -0.557196855545044,
|
| 672 |
+
"logits/rejected": -0.5299938321113586,
|
| 673 |
+
"logps/chosen": -159.45950317382812,
|
| 674 |
+
"logps/rejected": -168.58143615722656,
|
| 675 |
+
"loss": 1.4956,
|
| 676 |
+
"nll_loss": 0.4551734924316406,
|
| 677 |
+
"rewards/accuracies": 0.628125011920929,
|
| 678 |
+
"rewards/chosen": -15.945948600769043,
|
| 679 |
+
"rewards/margins": 0.9121924638748169,
|
| 680 |
+
"rewards/rejected": -16.858142852783203,
|
| 681 |
+
"step": 420
|
| 682 |
+
},
|
| 683 |
+
{
|
| 684 |
+
"epoch": 0.9939323894828085,
|
| 685 |
+
"grad_norm": 51.070491790771484,
|
| 686 |
+
"learning_rate": 7.427101200686106e-07,
|
| 687 |
+
"logits/chosen": -0.5334554314613342,
|
| 688 |
+
"logits/rejected": -0.5415940284729004,
|
| 689 |
+
"logps/chosen": -159.59457397460938,
|
| 690 |
+
"logps/rejected": -169.33242797851562,
|
| 691 |
+
"loss": 1.5867,
|
| 692 |
+
"nll_loss": 0.4326634407043457,
|
| 693 |
+
"rewards/accuracies": 0.609375,
|
| 694 |
+
"rewards/chosen": -15.95945930480957,
|
| 695 |
+
"rewards/margins": 0.973785400390625,
|
| 696 |
+
"rewards/rejected": -16.933242797851562,
|
| 697 |
+
"step": 430
|
| 698 |
+
},
|
| 699 |
+
{
|
| 700 |
+
"epoch": 0.9985553308292401,
|
| 701 |
+
"eval_logits/chosen": -0.5128109455108643,
|
| 702 |
+
"eval_logits/rejected": -0.478252649307251,
|
| 703 |
+
"eval_logps/chosen": -160.0940704345703,
|
| 704 |
+
"eval_logps/rejected": -169.7456817626953,
|
| 705 |
+
"eval_loss": 1.5247759819030762,
|
| 706 |
+
"eval_nll_loss": 0.43732327222824097,
|
| 707 |
+
"eval_rewards/accuracies": 0.658695638179779,
|
| 708 |
+
"eval_rewards/chosen": -16.00940704345703,
|
| 709 |
+
"eval_rewards/margins": 0.9651613831520081,
|
| 710 |
+
"eval_rewards/rejected": -16.97456932067871,
|
| 711 |
+
"eval_runtime": 74.3146,
|
| 712 |
+
"eval_samples_per_second": 24.571,
|
| 713 |
+
"eval_steps_per_second": 1.547,
|
| 714 |
+
"step": 432
|
| 715 |
+
},
|
| 716 |
+
{
|
| 717 |
+
"epoch": 1.0170470962149667,
|
| 718 |
+
"grad_norm": 31.19192886352539,
|
| 719 |
+
"learning_rate": 7.341337907375643e-07,
|
| 720 |
+
"logits/chosen": -0.5276386141777039,
|
| 721 |
+
"logits/rejected": -0.53285151720047,
|
| 722 |
+
"logps/chosen": -151.15855407714844,
|
| 723 |
+
"logps/rejected": -171.3574676513672,
|
| 724 |
+
"loss": 1.053,
|
| 725 |
+
"nll_loss": 0.41920948028564453,
|
| 726 |
+
"rewards/accuracies": 0.7906249761581421,
|
| 727 |
+
"rewards/chosen": -15.115857124328613,
|
| 728 |
+
"rewards/margins": 2.019890308380127,
|
| 729 |
+
"rewards/rejected": -17.135746002197266,
|
| 730 |
+
"step": 440
|
| 731 |
+
},
|
| 732 |
+
{
|
| 733 |
+
"epoch": 1.0401618029471251,
|
| 734 |
+
"grad_norm": 29.320512771606445,
|
| 735 |
+
"learning_rate": 7.255574614065179e-07,
|
| 736 |
+
"logits/chosen": -0.5006138682365417,
|
| 737 |
+
"logits/rejected": -0.4941863417625427,
|
| 738 |
+
"logps/chosen": -140.65113830566406,
|
| 739 |
+
"logps/rejected": -162.8360137939453,
|
| 740 |
+
"loss": 0.8732,
|
| 741 |
+
"nll_loss": 0.39865946769714355,
|
| 742 |
+
"rewards/accuracies": 0.7749999761581421,
|
| 743 |
+
"rewards/chosen": -14.065114974975586,
|
| 744 |
+
"rewards/margins": 2.2184863090515137,
|
| 745 |
+
"rewards/rejected": -16.283601760864258,
|
| 746 |
+
"step": 450
|
| 747 |
+
},
|
| 748 |
+
{
|
| 749 |
+
"epoch": 1.0632765096792833,
|
| 750 |
+
"grad_norm": 35.25569152832031,
|
| 751 |
+
"learning_rate": 7.169811320754716e-07,
|
| 752 |
+
"logits/chosen": -0.5818579196929932,
|
| 753 |
+
"logits/rejected": -0.560608983039856,
|
| 754 |
+
"logps/chosen": -144.07252502441406,
|
| 755 |
+
"logps/rejected": -166.63180541992188,
|
| 756 |
+
"loss": 0.8386,
|
| 757 |
+
"nll_loss": 0.3961290419101715,
|
| 758 |
+
"rewards/accuracies": 0.859375,
|
| 759 |
+
"rewards/chosen": -14.407251358032227,
|
| 760 |
+
"rewards/margins": 2.2559285163879395,
|
| 761 |
+
"rewards/rejected": -16.663179397583008,
|
| 762 |
+
"step": 460
|
| 763 |
+
},
|
| 764 |
+
{
|
| 765 |
+
"epoch": 1.0863912164114418,
|
| 766 |
+
"grad_norm": 43.183834075927734,
|
| 767 |
+
"learning_rate": 7.084048027444254e-07,
|
| 768 |
+
"logits/chosen": -0.5513390302658081,
|
| 769 |
+
"logits/rejected": -0.5131951570510864,
|
| 770 |
+
"logps/chosen": -143.40451049804688,
|
| 771 |
+
"logps/rejected": -168.33999633789062,
|
| 772 |
+
"loss": 0.8444,
|
| 773 |
+
"nll_loss": 0.3932550549507141,
|
| 774 |
+
"rewards/accuracies": 0.84375,
|
| 775 |
+
"rewards/chosen": -14.34045124053955,
|
| 776 |
+
"rewards/margins": 2.4935507774353027,
|
| 777 |
+
"rewards/rejected": -16.834003448486328,
|
| 778 |
+
"step": 470
|
| 779 |
+
},
|
| 780 |
+
{
|
| 781 |
+
"epoch": 1.1095059231436002,
|
| 782 |
+
"grad_norm": 46.28910446166992,
|
| 783 |
+
"learning_rate": 6.99828473413379e-07,
|
| 784 |
+
"logits/chosen": -0.558716893196106,
|
| 785 |
+
"logits/rejected": -0.5409826040267944,
|
| 786 |
+
"logps/chosen": -149.60780334472656,
|
| 787 |
+
"logps/rejected": -174.61407470703125,
|
| 788 |
+
"loss": 0.8528,
|
| 789 |
+
"nll_loss": 0.38355833292007446,
|
| 790 |
+
"rewards/accuracies": 0.8343750238418579,
|
| 791 |
+
"rewards/chosen": -14.960783004760742,
|
| 792 |
+
"rewards/margins": 2.5006256103515625,
|
| 793 |
+
"rewards/rejected": -17.461406707763672,
|
| 794 |
+
"step": 480
|
| 795 |
+
},
|
| 796 |
+
{
|
| 797 |
+
"epoch": 1.1326206298757584,
|
| 798 |
+
"grad_norm": 26.747591018676758,
|
| 799 |
+
"learning_rate": 6.912521440823327e-07,
|
| 800 |
+
"logits/chosen": -0.5055437684059143,
|
| 801 |
+
"logits/rejected": -0.48391538858413696,
|
| 802 |
+
"logps/chosen": -140.9654083251953,
|
| 803 |
+
"logps/rejected": -165.58120727539062,
|
| 804 |
+
"loss": 0.7887,
|
| 805 |
+
"nll_loss": 0.39576801657676697,
|
| 806 |
+
"rewards/accuracies": 0.8656250238418579,
|
| 807 |
+
"rewards/chosen": -14.096539497375488,
|
| 808 |
+
"rewards/margins": 2.4615824222564697,
|
| 809 |
+
"rewards/rejected": -16.558120727539062,
|
| 810 |
+
"step": 490
|
| 811 |
+
},
|
| 812 |
+
{
|
| 813 |
+
"epoch": 1.1557353366079168,
|
| 814 |
+
"grad_norm": 35.81715774536133,
|
| 815 |
+
"learning_rate": 6.826758147512864e-07,
|
| 816 |
+
"logits/chosen": -0.600662887096405,
|
| 817 |
+
"logits/rejected": -0.5748721361160278,
|
| 818 |
+
"logps/chosen": -140.01742553710938,
|
| 819 |
+
"logps/rejected": -164.94715881347656,
|
| 820 |
+
"loss": 0.8574,
|
| 821 |
+
"nll_loss": 0.38788530230522156,
|
| 822 |
+
"rewards/accuracies": 0.840624988079071,
|
| 823 |
+
"rewards/chosen": -14.001742362976074,
|
| 824 |
+
"rewards/margins": 2.492974042892456,
|
| 825 |
+
"rewards/rejected": -16.49471664428711,
|
| 826 |
+
"step": 500
|
| 827 |
+
},
|
| 828 |
+
{
|
| 829 |
+
"epoch": 1.178850043340075,
|
| 830 |
+
"grad_norm": 32.66586685180664,
|
| 831 |
+
"learning_rate": 6.740994854202401e-07,
|
| 832 |
+
"logits/chosen": -0.6161425113677979,
|
| 833 |
+
"logits/rejected": -0.6100048422813416,
|
| 834 |
+
"logps/chosen": -145.22962951660156,
|
| 835 |
+
"logps/rejected": -166.43936157226562,
|
| 836 |
+
"loss": 0.8359,
|
| 837 |
+
"nll_loss": 0.3928259015083313,
|
| 838 |
+
"rewards/accuracies": 0.8062499761581421,
|
| 839 |
+
"rewards/chosen": -14.52296257019043,
|
| 840 |
+
"rewards/margins": 2.120974063873291,
|
| 841 |
+
"rewards/rejected": -16.643936157226562,
|
| 842 |
+
"step": 510
|
| 843 |
+
},
|
| 844 |
+
{
|
| 845 |
+
"epoch": 1.2019647500722335,
|
| 846 |
+
"grad_norm": 33.49191665649414,
|
| 847 |
+
"learning_rate": 6.655231560891939e-07,
|
| 848 |
+
"logits/chosen": -0.5515680313110352,
|
| 849 |
+
"logits/rejected": -0.5309056043624878,
|
| 850 |
+
"logps/chosen": -149.71470642089844,
|
| 851 |
+
"logps/rejected": -176.19627380371094,
|
| 852 |
+
"loss": 0.8231,
|
| 853 |
+
"nll_loss": 0.4030747413635254,
|
| 854 |
+
"rewards/accuracies": 0.846875011920929,
|
| 855 |
+
"rewards/chosen": -14.971471786499023,
|
| 856 |
+
"rewards/margins": 2.6481575965881348,
|
| 857 |
+
"rewards/rejected": -17.619626998901367,
|
| 858 |
+
"step": 520
|
| 859 |
+
},
|
| 860 |
+
{
|
| 861 |
+
"epoch": 1.2250794568043917,
|
| 862 |
+
"grad_norm": 34.94106674194336,
|
| 863 |
+
"learning_rate": 6.569468267581475e-07,
|
| 864 |
+
"logits/chosen": -0.48874059319496155,
|
| 865 |
+
"logits/rejected": -0.4897806644439697,
|
| 866 |
+
"logps/chosen": -143.54364013671875,
|
| 867 |
+
"logps/rejected": -170.92356872558594,
|
| 868 |
+
"loss": 0.8061,
|
| 869 |
+
"nll_loss": 0.3903985321521759,
|
| 870 |
+
"rewards/accuracies": 0.8374999761581421,
|
| 871 |
+
"rewards/chosen": -14.354365348815918,
|
| 872 |
+
"rewards/margins": 2.7379937171936035,
|
| 873 |
+
"rewards/rejected": -17.092357635498047,
|
| 874 |
+
"step": 530
|
| 875 |
+
},
|
| 876 |
+
{
|
| 877 |
+
"epoch": 1.24819416353655,
|
| 878 |
+
"grad_norm": 38.41946792602539,
|
| 879 |
+
"learning_rate": 6.483704974271011e-07,
|
| 880 |
+
"logits/chosen": -0.5544147491455078,
|
| 881 |
+
"logits/rejected": -0.5470559000968933,
|
| 882 |
+
"logps/chosen": -143.85928344726562,
|
| 883 |
+
"logps/rejected": -165.72219848632812,
|
| 884 |
+
"loss": 0.8594,
|
| 885 |
+
"nll_loss": 0.4008564352989197,
|
| 886 |
+
"rewards/accuracies": 0.824999988079071,
|
| 887 |
+
"rewards/chosen": -14.385930061340332,
|
| 888 |
+
"rewards/margins": 2.1862895488739014,
|
| 889 |
+
"rewards/rejected": -16.572219848632812,
|
| 890 |
+
"step": 540
|
| 891 |
+
},
|
| 892 |
+
{
|
| 893 |
+
"epoch": 1.2713088702687085,
|
| 894 |
+
"grad_norm": 34.09145736694336,
|
| 895 |
+
"learning_rate": 6.397941680960549e-07,
|
| 896 |
+
"logits/chosen": -0.6739064455032349,
|
| 897 |
+
"logits/rejected": -0.6665552258491516,
|
| 898 |
+
"logps/chosen": -153.54989624023438,
|
| 899 |
+
"logps/rejected": -179.6954345703125,
|
| 900 |
+
"loss": 0.7542,
|
| 901 |
+
"nll_loss": 0.4033665060997009,
|
| 902 |
+
"rewards/accuracies": 0.8687499761581421,
|
| 903 |
+
"rewards/chosen": -15.354990005493164,
|
| 904 |
+
"rewards/margins": 2.6145541667938232,
|
| 905 |
+
"rewards/rejected": -17.96954345703125,
|
| 906 |
+
"step": 550
|
| 907 |
+
},
|
| 908 |
+
{
|
| 909 |
+
"epoch": 1.2944235770008667,
|
| 910 |
+
"grad_norm": 33.82338333129883,
|
| 911 |
+
"learning_rate": 6.312178387650086e-07,
|
| 912 |
+
"logits/chosen": -0.6586141586303711,
|
| 913 |
+
"logits/rejected": -0.6481257677078247,
|
| 914 |
+
"logps/chosen": -140.09243774414062,
|
| 915 |
+
"logps/rejected": -168.21730041503906,
|
| 916 |
+
"loss": 0.8033,
|
| 917 |
+
"nll_loss": 0.383733332157135,
|
| 918 |
+
"rewards/accuracies": 0.875,
|
| 919 |
+
"rewards/chosen": -14.009244918823242,
|
| 920 |
+
"rewards/margins": 2.812483310699463,
|
| 921 |
+
"rewards/rejected": -16.821727752685547,
|
| 922 |
+
"step": 560
|
| 923 |
+
},
|
| 924 |
+
{
|
| 925 |
+
"epoch": 1.3175382837330252,
|
| 926 |
+
"grad_norm": 37.8900032043457,
|
| 927 |
+
"learning_rate": 6.226415094339622e-07,
|
| 928 |
+
"logits/chosen": -0.5938795804977417,
|
| 929 |
+
"logits/rejected": -0.5684852600097656,
|
| 930 |
+
"logps/chosen": -150.65396118164062,
|
| 931 |
+
"logps/rejected": -176.8959503173828,
|
| 932 |
+
"loss": 0.8343,
|
| 933 |
+
"nll_loss": 0.40136751532554626,
|
| 934 |
+
"rewards/accuracies": 0.859375,
|
| 935 |
+
"rewards/chosen": -15.065396308898926,
|
| 936 |
+
"rewards/margins": 2.624197244644165,
|
| 937 |
+
"rewards/rejected": -17.689594268798828,
|
| 938 |
+
"step": 570
|
| 939 |
+
},
|
| 940 |
+
{
|
| 941 |
+
"epoch": 1.3406529904651836,
|
| 942 |
+
"grad_norm": 41.33928298950195,
|
| 943 |
+
"learning_rate": 6.14065180102916e-07,
|
| 944 |
+
"logits/chosen": -0.5615830421447754,
|
| 945 |
+
"logits/rejected": -0.5374659299850464,
|
| 946 |
+
"logps/chosen": -146.83155822753906,
|
| 947 |
+
"logps/rejected": -174.39060974121094,
|
| 948 |
+
"loss": 0.8152,
|
| 949 |
+
"nll_loss": 0.3893912732601166,
|
| 950 |
+
"rewards/accuracies": 0.859375,
|
| 951 |
+
"rewards/chosen": -14.683156967163086,
|
| 952 |
+
"rewards/margins": 2.7559053897857666,
|
| 953 |
+
"rewards/rejected": -17.43906021118164,
|
| 954 |
+
"step": 580
|
| 955 |
+
},
|
| 956 |
+
{
|
| 957 |
+
"epoch": 1.3637676971973418,
|
| 958 |
+
"grad_norm": 40.57521438598633,
|
| 959 |
+
"learning_rate": 6.054888507718696e-07,
|
| 960 |
+
"logits/chosen": -0.5162426233291626,
|
| 961 |
+
"logits/rejected": -0.5248137712478638,
|
| 962 |
+
"logps/chosen": -140.13807678222656,
|
| 963 |
+
"logps/rejected": -166.10623168945312,
|
| 964 |
+
"loss": 0.7793,
|
| 965 |
+
"nll_loss": 0.4015275537967682,
|
| 966 |
+
"rewards/accuracies": 0.840624988079071,
|
| 967 |
+
"rewards/chosen": -14.013806343078613,
|
| 968 |
+
"rewards/margins": 2.596815347671509,
|
| 969 |
+
"rewards/rejected": -16.61062240600586,
|
| 970 |
+
"step": 590
|
| 971 |
+
},
|
| 972 |
+
{
|
| 973 |
+
"epoch": 1.3868824039295,
|
| 974 |
+
"grad_norm": 32.84572219848633,
|
| 975 |
+
"learning_rate": 5.969125214408233e-07,
|
| 976 |
+
"logits/chosen": -0.5050234198570251,
|
| 977 |
+
"logits/rejected": -0.4766197204589844,
|
| 978 |
+
"logps/chosen": -143.4470672607422,
|
| 979 |
+
"logps/rejected": -169.41189575195312,
|
| 980 |
+
"loss": 0.8246,
|
| 981 |
+
"nll_loss": 0.377535343170166,
|
| 982 |
+
"rewards/accuracies": 0.831250011920929,
|
| 983 |
+
"rewards/chosen": -14.344708442687988,
|
| 984 |
+
"rewards/margins": 2.5964834690093994,
|
| 985 |
+
"rewards/rejected": -16.941190719604492,
|
| 986 |
+
"step": 600
|
| 987 |
+
},
|
| 988 |
+
{
|
| 989 |
+
"epoch": 1.4099971106616584,
|
| 990 |
+
"grad_norm": 35.82289123535156,
|
| 991 |
+
"learning_rate": 5.88336192109777e-07,
|
| 992 |
+
"logits/chosen": -0.477673202753067,
|
| 993 |
+
"logits/rejected": -0.4712616801261902,
|
| 994 |
+
"logps/chosen": -141.12518310546875,
|
| 995 |
+
"logps/rejected": -165.6627655029297,
|
| 996 |
+
"loss": 0.8663,
|
| 997 |
+
"nll_loss": 0.40321165323257446,
|
| 998 |
+
"rewards/accuracies": 0.846875011920929,
|
| 999 |
+
"rewards/chosen": -14.112518310546875,
|
| 1000 |
+
"rewards/margins": 2.453758716583252,
|
| 1001 |
+
"rewards/rejected": -16.5662784576416,
|
| 1002 |
+
"step": 610
|
| 1003 |
+
},
|
| 1004 |
+
{
|
| 1005 |
+
"epoch": 1.4331118173938169,
|
| 1006 |
+
"grad_norm": 22.42660903930664,
|
| 1007 |
+
"learning_rate": 5.797598627787307e-07,
|
| 1008 |
+
"logits/chosen": -0.559418797492981,
|
| 1009 |
+
"logits/rejected": -0.5618263483047485,
|
| 1010 |
+
"logps/chosen": -147.1465301513672,
|
| 1011 |
+
"logps/rejected": -175.36915588378906,
|
| 1012 |
+
"loss": 0.718,
|
| 1013 |
+
"nll_loss": 0.3817431330680847,
|
| 1014 |
+
"rewards/accuracies": 0.878125011920929,
|
| 1015 |
+
"rewards/chosen": -14.714654922485352,
|
| 1016 |
+
"rewards/margins": 2.8222622871398926,
|
| 1017 |
+
"rewards/rejected": -17.536914825439453,
|
| 1018 |
+
"step": 620
|
| 1019 |
+
},
|
| 1020 |
+
{
|
| 1021 |
+
"epoch": 1.456226524125975,
|
| 1022 |
+
"grad_norm": 36.228431701660156,
|
| 1023 |
+
"learning_rate": 5.711835334476843e-07,
|
| 1024 |
+
"logits/chosen": -0.45862525701522827,
|
| 1025 |
+
"logits/rejected": -0.4514252245426178,
|
| 1026 |
+
"logps/chosen": -134.56240844726562,
|
| 1027 |
+
"logps/rejected": -161.8486328125,
|
| 1028 |
+
"loss": 0.757,
|
| 1029 |
+
"nll_loss": 0.37470743060112,
|
| 1030 |
+
"rewards/accuracies": 0.846875011920929,
|
| 1031 |
+
"rewards/chosen": -13.4562406539917,
|
| 1032 |
+
"rewards/margins": 2.7286224365234375,
|
| 1033 |
+
"rewards/rejected": -16.184864044189453,
|
| 1034 |
+
"step": 630
|
| 1035 |
+
},
|
| 1036 |
+
{
|
| 1037 |
+
"epoch": 1.4793412308581335,
|
| 1038 |
+
"grad_norm": 37.52621078491211,
|
| 1039 |
+
"learning_rate": 5.626072041166381e-07,
|
| 1040 |
+
"logits/chosen": -0.4572775363922119,
|
| 1041 |
+
"logits/rejected": -0.4576060175895691,
|
| 1042 |
+
"logps/chosen": -132.4957275390625,
|
| 1043 |
+
"logps/rejected": -160.1851348876953,
|
| 1044 |
+
"loss": 0.8095,
|
| 1045 |
+
"nll_loss": 0.38259443640708923,
|
| 1046 |
+
"rewards/accuracies": 0.8843749761581421,
|
| 1047 |
+
"rewards/chosen": -13.249574661254883,
|
| 1048 |
+
"rewards/margins": 2.768939733505249,
|
| 1049 |
+
"rewards/rejected": -16.018512725830078,
|
| 1050 |
+
"step": 640
|
| 1051 |
+
},
|
| 1052 |
+
{
|
| 1053 |
+
"epoch": 1.502455937590292,
|
| 1054 |
+
"grad_norm": 31.980655670166016,
|
| 1055 |
+
"learning_rate": 5.540308747855917e-07,
|
| 1056 |
+
"logits/chosen": -0.514967143535614,
|
| 1057 |
+
"logits/rejected": -0.5276812314987183,
|
| 1058 |
+
"logps/chosen": -144.45018005371094,
|
| 1059 |
+
"logps/rejected": -174.40380859375,
|
| 1060 |
+
"loss": 0.7572,
|
| 1061 |
+
"nll_loss": 0.391319215297699,
|
| 1062 |
+
"rewards/accuracies": 0.893750011920929,
|
| 1063 |
+
"rewards/chosen": -14.445019721984863,
|
| 1064 |
+
"rewards/margins": 2.995361328125,
|
| 1065 |
+
"rewards/rejected": -17.440380096435547,
|
| 1066 |
+
"step": 650
|
| 1067 |
+
},
|
| 1068 |
+
{
|
| 1069 |
+
"epoch": 1.5255706443224502,
|
| 1070 |
+
"grad_norm": 33.730308532714844,
|
| 1071 |
+
"learning_rate": 5.454545454545454e-07,
|
| 1072 |
+
"logits/chosen": -0.4829156994819641,
|
| 1073 |
+
"logits/rejected": -0.4813632071018219,
|
| 1074 |
+
"logps/chosen": -136.35617065429688,
|
| 1075 |
+
"logps/rejected": -163.93539428710938,
|
| 1076 |
+
"loss": 0.8365,
|
| 1077 |
+
"nll_loss": 0.3918777108192444,
|
| 1078 |
+
"rewards/accuracies": 0.828125,
|
| 1079 |
+
"rewards/chosen": -13.63561725616455,
|
| 1080 |
+
"rewards/margins": 2.757922887802124,
|
| 1081 |
+
"rewards/rejected": -16.393539428710938,
|
| 1082 |
+
"step": 660
|
| 1083 |
+
},
|
| 1084 |
+
{
|
| 1085 |
+
"epoch": 1.5486853510546084,
|
| 1086 |
+
"grad_norm": 37.20032501220703,
|
| 1087 |
+
"learning_rate": 5.368782161234992e-07,
|
| 1088 |
+
"logits/chosen": -0.5203684568405151,
|
| 1089 |
+
"logits/rejected": -0.5274239778518677,
|
| 1090 |
+
"logps/chosen": -147.5542449951172,
|
| 1091 |
+
"logps/rejected": -175.27745056152344,
|
| 1092 |
+
"loss": 0.7471,
|
| 1093 |
+
"nll_loss": 0.3757604956626892,
|
| 1094 |
+
"rewards/accuracies": 0.8531249761581421,
|
| 1095 |
+
"rewards/chosen": -14.755424499511719,
|
| 1096 |
+
"rewards/margins": 2.772320032119751,
|
| 1097 |
+
"rewards/rejected": -17.527746200561523,
|
| 1098 |
+
"step": 670
|
| 1099 |
+
},
|
| 1100 |
+
{
|
| 1101 |
+
"epoch": 1.5718000577867668,
|
| 1102 |
+
"grad_norm": 30.996845245361328,
|
| 1103 |
+
"learning_rate": 5.283018867924528e-07,
|
| 1104 |
+
"logits/chosen": -0.4579756259918213,
|
| 1105 |
+
"logits/rejected": -0.46921706199645996,
|
| 1106 |
+
"logps/chosen": -141.15423583984375,
|
| 1107 |
+
"logps/rejected": -167.1835479736328,
|
| 1108 |
+
"loss": 0.8266,
|
| 1109 |
+
"nll_loss": 0.3890347182750702,
|
| 1110 |
+
"rewards/accuracies": 0.8343750238418579,
|
| 1111 |
+
"rewards/chosen": -14.115423202514648,
|
| 1112 |
+
"rewards/margins": 2.6029300689697266,
|
| 1113 |
+
"rewards/rejected": -16.718355178833008,
|
| 1114 |
+
"step": 680
|
| 1115 |
+
},
|
| 1116 |
+
{
|
| 1117 |
+
"epoch": 1.5949147645189252,
|
| 1118 |
+
"grad_norm": 24.3118896484375,
|
| 1119 |
+
"learning_rate": 5.197255574614064e-07,
|
| 1120 |
+
"logits/chosen": -0.4211729168891907,
|
| 1121 |
+
"logits/rejected": -0.42907238006591797,
|
| 1122 |
+
"logps/chosen": -149.1224365234375,
|
| 1123 |
+
"logps/rejected": -180.13198852539062,
|
| 1124 |
+
"loss": 0.7815,
|
| 1125 |
+
"nll_loss": 0.38901641964912415,
|
| 1126 |
+
"rewards/accuracies": 0.856249988079071,
|
| 1127 |
+
"rewards/chosen": -14.91224479675293,
|
| 1128 |
+
"rewards/margins": 3.100955009460449,
|
| 1129 |
+
"rewards/rejected": -18.013200759887695,
|
| 1130 |
+
"step": 690
|
| 1131 |
+
},
|
| 1132 |
+
{
|
| 1133 |
+
"epoch": 1.6180294712510834,
|
| 1134 |
+
"grad_norm": 31.886476516723633,
|
| 1135 |
+
"learning_rate": 5.111492281303602e-07,
|
| 1136 |
+
"logits/chosen": -0.36691445112228394,
|
| 1137 |
+
"logits/rejected": -0.33470767736434937,
|
| 1138 |
+
"logps/chosen": -139.63748168945312,
|
| 1139 |
+
"logps/rejected": -166.2130889892578,
|
| 1140 |
+
"loss": 0.7878,
|
| 1141 |
+
"nll_loss": 0.38116249442100525,
|
| 1142 |
+
"rewards/accuracies": 0.84375,
|
| 1143 |
+
"rewards/chosen": -13.96374797821045,
|
| 1144 |
+
"rewards/margins": 2.657560110092163,
|
| 1145 |
+
"rewards/rejected": -16.621309280395508,
|
| 1146 |
+
"step": 700
|
| 1147 |
+
},
|
| 1148 |
+
{
|
| 1149 |
+
"epoch": 1.6411441779832419,
|
| 1150 |
+
"grad_norm": 26.3208065032959,
|
| 1151 |
+
"learning_rate": 5.025728987993139e-07,
|
| 1152 |
+
"logits/chosen": -0.3681151866912842,
|
| 1153 |
+
"logits/rejected": -0.3645099103450775,
|
| 1154 |
+
"logps/chosen": -127.74934387207031,
|
| 1155 |
+
"logps/rejected": -153.67477416992188,
|
| 1156 |
+
"loss": 0.7603,
|
| 1157 |
+
"nll_loss": 0.363952100276947,
|
| 1158 |
+
"rewards/accuracies": 0.8687499761581421,
|
| 1159 |
+
"rewards/chosen": -12.774932861328125,
|
| 1160 |
+
"rewards/margins": 2.5925447940826416,
|
| 1161 |
+
"rewards/rejected": -15.367477416992188,
|
| 1162 |
+
"step": 710
|
| 1163 |
+
},
|
| 1164 |
+
{
|
| 1165 |
+
"epoch": 1.6642588847154003,
|
| 1166 |
+
"grad_norm": 34.93773651123047,
|
| 1167 |
+
"learning_rate": 4.939965694682676e-07,
|
| 1168 |
+
"logits/chosen": -0.4657221734523773,
|
| 1169 |
+
"logits/rejected": -0.47038349509239197,
|
| 1170 |
+
"logps/chosen": -130.09898376464844,
|
| 1171 |
+
"logps/rejected": -153.72354125976562,
|
| 1172 |
+
"loss": 0.7346,
|
| 1173 |
+
"nll_loss": 0.391609787940979,
|
| 1174 |
+
"rewards/accuracies": 0.846875011920929,
|
| 1175 |
+
"rewards/chosen": -13.00989818572998,
|
| 1176 |
+
"rewards/margins": 2.362457513809204,
|
| 1177 |
+
"rewards/rejected": -15.372354507446289,
|
| 1178 |
+
"step": 720
|
| 1179 |
+
},
|
| 1180 |
+
{
|
| 1181 |
+
"epoch": 1.6873735914475585,
|
| 1182 |
+
"grad_norm": 32.160396575927734,
|
| 1183 |
+
"learning_rate": 4.854202401372212e-07,
|
| 1184 |
+
"logits/chosen": -0.49930065870285034,
|
| 1185 |
+
"logits/rejected": -0.4945829510688782,
|
| 1186 |
+
"logps/chosen": -139.16799926757812,
|
| 1187 |
+
"logps/rejected": -165.69920349121094,
|
| 1188 |
+
"loss": 0.7889,
|
| 1189 |
+
"nll_loss": 0.3781120777130127,
|
| 1190 |
+
"rewards/accuracies": 0.862500011920929,
|
| 1191 |
+
"rewards/chosen": -13.916799545288086,
|
| 1192 |
+
"rewards/margins": 2.6531200408935547,
|
| 1193 |
+
"rewards/rejected": -16.569921493530273,
|
| 1194 |
+
"step": 730
|
| 1195 |
+
},
|
| 1196 |
+
{
|
| 1197 |
+
"epoch": 1.7104882981797167,
|
| 1198 |
+
"grad_norm": 32.3018798828125,
|
| 1199 |
+
"learning_rate": 4.768439108061749e-07,
|
| 1200 |
+
"logits/chosen": -0.43461236357688904,
|
| 1201 |
+
"logits/rejected": -0.43381839990615845,
|
| 1202 |
+
"logps/chosen": -149.93421936035156,
|
| 1203 |
+
"logps/rejected": -180.68209838867188,
|
| 1204 |
+
"loss": 0.7308,
|
| 1205 |
+
"nll_loss": 0.4099641740322113,
|
| 1206 |
+
"rewards/accuracies": 0.8656250238418579,
|
| 1207 |
+
"rewards/chosen": -14.99342155456543,
|
| 1208 |
+
"rewards/margins": 3.074786901473999,
|
| 1209 |
+
"rewards/rejected": -18.06821060180664,
|
| 1210 |
+
"step": 740
|
| 1211 |
+
},
|
| 1212 |
+
{
|
| 1213 |
+
"epoch": 1.7336030049118751,
|
| 1214 |
+
"grad_norm": 36.8819694519043,
|
| 1215 |
+
"learning_rate": 4.6826758147512864e-07,
|
| 1216 |
+
"logits/chosen": -0.4040675759315491,
|
| 1217 |
+
"logits/rejected": -0.3790331482887268,
|
| 1218 |
+
"logps/chosen": -129.85499572753906,
|
| 1219 |
+
"logps/rejected": -156.1541290283203,
|
| 1220 |
+
"loss": 0.7651,
|
| 1221 |
+
"nll_loss": 0.3684754967689514,
|
| 1222 |
+
"rewards/accuracies": 0.840624988079071,
|
| 1223 |
+
"rewards/chosen": -12.985501289367676,
|
| 1224 |
+
"rewards/margins": 2.629912853240967,
|
| 1225 |
+
"rewards/rejected": -15.615412712097168,
|
| 1226 |
+
"step": 750
|
| 1227 |
+
},
|
| 1228 |
+
{
|
| 1229 |
+
"epoch": 1.7567177116440336,
|
| 1230 |
+
"grad_norm": 28.750247955322266,
|
| 1231 |
+
"learning_rate": 4.596912521440823e-07,
|
| 1232 |
+
"logits/chosen": -0.35753822326660156,
|
| 1233 |
+
"logits/rejected": -0.3664765954017639,
|
| 1234 |
+
"logps/chosen": -141.37014770507812,
|
| 1235 |
+
"logps/rejected": -173.26290893554688,
|
| 1236 |
+
"loss": 0.7928,
|
| 1237 |
+
"nll_loss": 0.37667417526245117,
|
| 1238 |
+
"rewards/accuracies": 0.8687499761581421,
|
| 1239 |
+
"rewards/chosen": -14.13701343536377,
|
| 1240 |
+
"rewards/margins": 3.1892781257629395,
|
| 1241 |
+
"rewards/rejected": -17.326290130615234,
|
| 1242 |
+
"step": 760
|
| 1243 |
+
},
|
| 1244 |
+
{
|
| 1245 |
+
"epoch": 1.7798324183761918,
|
| 1246 |
+
"grad_norm": 31.120779037475586,
|
| 1247 |
+
"learning_rate": 4.51114922813036e-07,
|
| 1248 |
+
"logits/chosen": -0.3904396593570709,
|
| 1249 |
+
"logits/rejected": -0.37857958674430847,
|
| 1250 |
+
"logps/chosen": -134.3868408203125,
|
| 1251 |
+
"logps/rejected": -159.94497680664062,
|
| 1252 |
+
"loss": 0.7942,
|
| 1253 |
+
"nll_loss": 0.38718339800834656,
|
| 1254 |
+
"rewards/accuracies": 0.815625011920929,
|
| 1255 |
+
"rewards/chosen": -13.438684463500977,
|
| 1256 |
+
"rewards/margins": 2.5558130741119385,
|
| 1257 |
+
"rewards/rejected": -15.994497299194336,
|
| 1258 |
+
"step": 770
|
| 1259 |
+
},
|
| 1260 |
+
{
|
| 1261 |
+
"epoch": 1.8029471251083502,
|
| 1262 |
+
"grad_norm": 30.371490478515625,
|
| 1263 |
+
"learning_rate": 4.4253859348198967e-07,
|
| 1264 |
+
"logits/chosen": -0.3549109101295471,
|
| 1265 |
+
"logits/rejected": -0.3400852680206299,
|
| 1266 |
+
"logps/chosen": -137.1378631591797,
|
| 1267 |
+
"logps/rejected": -166.12417602539062,
|
| 1268 |
+
"loss": 0.7521,
|
| 1269 |
+
"nll_loss": 0.38493841886520386,
|
| 1270 |
+
"rewards/accuracies": 0.893750011920929,
|
| 1271 |
+
"rewards/chosen": -13.713786125183105,
|
| 1272 |
+
"rewards/margins": 2.898630380630493,
|
| 1273 |
+
"rewards/rejected": -16.612417221069336,
|
| 1274 |
+
"step": 780
|
| 1275 |
+
},
|
| 1276 |
+
{
|
| 1277 |
+
"epoch": 1.8260618318405086,
|
| 1278 |
+
"grad_norm": 35.5169677734375,
|
| 1279 |
+
"learning_rate": 4.339622641509434e-07,
|
| 1280 |
+
"logits/chosen": -0.3591529428958893,
|
| 1281 |
+
"logits/rejected": -0.3554760813713074,
|
| 1282 |
+
"logps/chosen": -138.97694396972656,
|
| 1283 |
+
"logps/rejected": -165.46890258789062,
|
| 1284 |
+
"loss": 0.7557,
|
| 1285 |
+
"nll_loss": 0.38407889008522034,
|
| 1286 |
+
"rewards/accuracies": 0.8531249761581421,
|
| 1287 |
+
"rewards/chosen": -13.89769458770752,
|
| 1288 |
+
"rewards/margins": 2.6491942405700684,
|
| 1289 |
+
"rewards/rejected": -16.546890258789062,
|
| 1290 |
+
"step": 790
|
| 1291 |
+
},
|
| 1292 |
+
{
|
| 1293 |
+
"epoch": 1.8491765385726668,
|
| 1294 |
+
"grad_norm": 32.5655403137207,
|
| 1295 |
+
"learning_rate": 4.25385934819897e-07,
|
| 1296 |
+
"logits/chosen": -0.2939130961894989,
|
| 1297 |
+
"logits/rejected": -0.2781139016151428,
|
| 1298 |
+
"logps/chosen": -127.32011413574219,
|
| 1299 |
+
"logps/rejected": -156.74063110351562,
|
| 1300 |
+
"loss": 0.7463,
|
| 1301 |
+
"nll_loss": 0.36602723598480225,
|
| 1302 |
+
"rewards/accuracies": 0.859375,
|
| 1303 |
+
"rewards/chosen": -12.732012748718262,
|
| 1304 |
+
"rewards/margins": 2.942049741744995,
|
| 1305 |
+
"rewards/rejected": -15.67406177520752,
|
| 1306 |
+
"step": 800
|
| 1307 |
+
},
|
| 1308 |
+
{
|
| 1309 |
+
"epoch": 1.872291245304825,
|
| 1310 |
+
"grad_norm": 31.026138305664062,
|
| 1311 |
+
"learning_rate": 4.1680960548885075e-07,
|
| 1312 |
+
"logits/chosen": -0.40986233949661255,
|
| 1313 |
+
"logits/rejected": -0.4055134356021881,
|
| 1314 |
+
"logps/chosen": -138.88014221191406,
|
| 1315 |
+
"logps/rejected": -168.54603576660156,
|
| 1316 |
+
"loss": 0.7996,
|
| 1317 |
+
"nll_loss": 0.38212329149246216,
|
| 1318 |
+
"rewards/accuracies": 0.8812500238418579,
|
| 1319 |
+
"rewards/chosen": -13.88801383972168,
|
| 1320 |
+
"rewards/margins": 2.9665894508361816,
|
| 1321 |
+
"rewards/rejected": -16.854602813720703,
|
| 1322 |
+
"step": 810
|
| 1323 |
+
},
|
| 1324 |
+
{
|
| 1325 |
+
"epoch": 1.8954059520369835,
|
| 1326 |
+
"grad_norm": 28.253450393676758,
|
| 1327 |
+
"learning_rate": 4.0823327615780443e-07,
|
| 1328 |
+
"logits/chosen": -0.42939743399620056,
|
| 1329 |
+
"logits/rejected": -0.42729324102401733,
|
| 1330 |
+
"logps/chosen": -138.198486328125,
|
| 1331 |
+
"logps/rejected": -170.29718017578125,
|
| 1332 |
+
"loss": 0.7528,
|
| 1333 |
+
"nll_loss": 0.38865524530410767,
|
| 1334 |
+
"rewards/accuracies": 0.871874988079071,
|
| 1335 |
+
"rewards/chosen": -13.819849967956543,
|
| 1336 |
+
"rewards/margins": 3.209869384765625,
|
| 1337 |
+
"rewards/rejected": -17.02971839904785,
|
| 1338 |
+
"step": 820
|
| 1339 |
+
},
|
| 1340 |
+
{
|
| 1341 |
+
"epoch": 1.918520658769142,
|
| 1342 |
+
"grad_norm": 38.18342590332031,
|
| 1343 |
+
"learning_rate": 3.9965694682675816e-07,
|
| 1344 |
+
"logits/chosen": -0.4124279022216797,
|
| 1345 |
+
"logits/rejected": -0.3961451053619385,
|
| 1346 |
+
"logps/chosen": -133.0750274658203,
|
| 1347 |
+
"logps/rejected": -158.16928100585938,
|
| 1348 |
+
"loss": 0.7746,
|
| 1349 |
+
"nll_loss": 0.39231568574905396,
|
| 1350 |
+
"rewards/accuracies": 0.84375,
|
| 1351 |
+
"rewards/chosen": -13.307504653930664,
|
| 1352 |
+
"rewards/margins": 2.509425401687622,
|
| 1353 |
+
"rewards/rejected": -15.816927909851074,
|
| 1354 |
+
"step": 830
|
| 1355 |
+
},
|
| 1356 |
+
{
|
| 1357 |
+
"epoch": 1.9416353655013001,
|
| 1358 |
+
"grad_norm": 33.664310455322266,
|
| 1359 |
+
"learning_rate": 3.910806174957118e-07,
|
| 1360 |
+
"logits/chosen": -0.5193787217140198,
|
| 1361 |
+
"logits/rejected": -0.5243502855300903,
|
| 1362 |
+
"logps/chosen": -139.03890991210938,
|
| 1363 |
+
"logps/rejected": -166.02572631835938,
|
| 1364 |
+
"loss": 0.7765,
|
| 1365 |
+
"nll_loss": 0.38118356466293335,
|
| 1366 |
+
"rewards/accuracies": 0.846875011920929,
|
| 1367 |
+
"rewards/chosen": -13.903889656066895,
|
| 1368 |
+
"rewards/margins": 2.6986842155456543,
|
| 1369 |
+
"rewards/rejected": -16.60257339477539,
|
| 1370 |
+
"step": 840
|
| 1371 |
+
},
|
| 1372 |
+
{
|
| 1373 |
+
"epoch": 1.9647500722334585,
|
| 1374 |
+
"grad_norm": 33.969181060791016,
|
| 1375 |
+
"learning_rate": 3.825042881646655e-07,
|
| 1376 |
+
"logits/chosen": -0.544953465461731,
|
| 1377 |
+
"logits/rejected": -0.5464820861816406,
|
| 1378 |
+
"logps/chosen": -138.60324096679688,
|
| 1379 |
+
"logps/rejected": -165.5134735107422,
|
| 1380 |
+
"loss": 0.755,
|
| 1381 |
+
"nll_loss": 0.37446194887161255,
|
| 1382 |
+
"rewards/accuracies": 0.90625,
|
| 1383 |
+
"rewards/chosen": -13.860323905944824,
|
| 1384 |
+
"rewards/margins": 2.6910228729248047,
|
| 1385 |
+
"rewards/rejected": -16.551347732543945,
|
| 1386 |
+
"step": 850
|
| 1387 |
+
},
|
| 1388 |
+
{
|
| 1389 |
+
"epoch": 1.987864778965617,
|
| 1390 |
+
"grad_norm": 33.37995529174805,
|
| 1391 |
+
"learning_rate": 3.739279588336192e-07,
|
| 1392 |
+
"logits/chosen": -0.5172384977340698,
|
| 1393 |
+
"logits/rejected": -0.5084408521652222,
|
| 1394 |
+
"logps/chosen": -138.13735961914062,
|
| 1395 |
+
"logps/rejected": -167.2960968017578,
|
| 1396 |
+
"loss": 0.7108,
|
| 1397 |
+
"nll_loss": 0.37610048055648804,
|
| 1398 |
+
"rewards/accuracies": 0.8687499761581421,
|
| 1399 |
+
"rewards/chosen": -13.813735961914062,
|
| 1400 |
+
"rewards/margins": 2.915872097015381,
|
| 1401 |
+
"rewards/rejected": -16.729610443115234,
|
| 1402 |
+
"step": 860
|
| 1403 |
+
},
|
| 1404 |
+
{
|
| 1405 |
+
"epoch": 1.999422132331696,
|
| 1406 |
+
"eval_logits/chosen": -0.46843183040618896,
|
| 1407 |
+
"eval_logits/rejected": -0.4402734041213989,
|
| 1408 |
+
"eval_logps/chosen": -148.37490844726562,
|
| 1409 |
+
"eval_logps/rejected": -159.45883178710938,
|
| 1410 |
+
"eval_loss": 1.5252443552017212,
|
| 1411 |
+
"eval_nll_loss": 0.4055543839931488,
|
| 1412 |
+
"eval_rewards/accuracies": 0.6499999761581421,
|
| 1413 |
+
"eval_rewards/chosen": -14.837491035461426,
|
| 1414 |
+
"eval_rewards/margins": 1.1083911657333374,
|
| 1415 |
+
"eval_rewards/rejected": -15.945883750915527,
|
| 1416 |
+
"eval_runtime": 74.3854,
|
| 1417 |
+
"eval_samples_per_second": 24.548,
|
| 1418 |
+
"eval_steps_per_second": 1.546,
|
| 1419 |
+
"step": 865
|
| 1420 |
+
},
|
| 1421 |
+
{
|
| 1422 |
+
"epoch": 2.010979485697775,
|
| 1423 |
+
"grad_norm": 13.765230178833008,
|
| 1424 |
+
"learning_rate": 3.6535162950257287e-07,
|
| 1425 |
+
"logits/chosen": -0.5043296217918396,
|
| 1426 |
+
"logits/rejected": -0.46686577796936035,
|
| 1427 |
+
"logps/chosen": -133.58786010742188,
|
| 1428 |
+
"logps/rejected": -166.25955200195312,
|
| 1429 |
+
"loss": 0.6206,
|
| 1430 |
+
"nll_loss": 0.35893407464027405,
|
| 1431 |
+
"rewards/accuracies": 0.903124988079071,
|
| 1432 |
+
"rewards/chosen": -13.358787536621094,
|
| 1433 |
+
"rewards/margins": 3.26716947555542,
|
| 1434 |
+
"rewards/rejected": -16.62595558166504,
|
| 1435 |
+
"step": 870
|
| 1436 |
+
},
|
| 1437 |
+
{
|
| 1438 |
+
"epoch": 2.0340941924299334,
|
| 1439 |
+
"grad_norm": 11.040721893310547,
|
| 1440 |
+
"learning_rate": 3.5677530017152655e-07,
|
| 1441 |
+
"logits/chosen": -0.44609642028808594,
|
| 1442 |
+
"logits/rejected": -0.45637327432632446,
|
| 1443 |
+
"logps/chosen": -124.71421813964844,
|
| 1444 |
+
"logps/rejected": -165.182373046875,
|
| 1445 |
+
"loss": 0.4779,
|
| 1446 |
+
"nll_loss": 0.35270074009895325,
|
| 1447 |
+
"rewards/accuracies": 0.971875011920929,
|
| 1448 |
+
"rewards/chosen": -12.47142219543457,
|
| 1449 |
+
"rewards/margins": 4.046815395355225,
|
| 1450 |
+
"rewards/rejected": -16.51823616027832,
|
| 1451 |
+
"step": 880
|
| 1452 |
+
},
|
| 1453 |
+
{
|
| 1454 |
+
"epoch": 2.057208899162092,
|
| 1455 |
+
"grad_norm": 15.88128662109375,
|
| 1456 |
+
"learning_rate": 3.481989708404803e-07,
|
| 1457 |
+
"logits/chosen": -0.3785150647163391,
|
| 1458 |
+
"logits/rejected": -0.3858146667480469,
|
| 1459 |
+
"logps/chosen": -123.47149658203125,
|
| 1460 |
+
"logps/rejected": -166.243408203125,
|
| 1461 |
+
"loss": 0.464,
|
| 1462 |
+
"nll_loss": 0.34830960631370544,
|
| 1463 |
+
"rewards/accuracies": 0.96875,
|
| 1464 |
+
"rewards/chosen": -12.347149848937988,
|
| 1465 |
+
"rewards/margins": 4.277192115783691,
|
| 1466 |
+
"rewards/rejected": -16.624340057373047,
|
| 1467 |
+
"step": 890
|
| 1468 |
+
},
|
| 1469 |
+
{
|
| 1470 |
+
"epoch": 2.0803236058942502,
|
| 1471 |
+
"grad_norm": 13.753726959228516,
|
| 1472 |
+
"learning_rate": 3.396226415094339e-07,
|
| 1473 |
+
"logits/chosen": -0.5063196420669556,
|
| 1474 |
+
"logits/rejected": -0.4997187554836273,
|
| 1475 |
+
"logps/chosen": -127.51603698730469,
|
| 1476 |
+
"logps/rejected": -165.35275268554688,
|
| 1477 |
+
"loss": 0.4607,
|
| 1478 |
+
"nll_loss": 0.3605991005897522,
|
| 1479 |
+
"rewards/accuracies": 0.987500011920929,
|
| 1480 |
+
"rewards/chosen": -12.751605033874512,
|
| 1481 |
+
"rewards/margins": 3.7836709022521973,
|
| 1482 |
+
"rewards/rejected": -16.535274505615234,
|
| 1483 |
+
"step": 900
|
| 1484 |
+
},
|
| 1485 |
+
{
|
| 1486 |
+
"epoch": 2.1034383126264085,
|
| 1487 |
+
"grad_norm": 16.0849666595459,
|
| 1488 |
+
"learning_rate": 3.3104631217838763e-07,
|
| 1489 |
+
"logits/chosen": -0.4099333882331848,
|
| 1490 |
+
"logits/rejected": -0.4197494387626648,
|
| 1491 |
+
"logps/chosen": -124.4246597290039,
|
| 1492 |
+
"logps/rejected": -165.6961669921875,
|
| 1493 |
+
"loss": 0.4726,
|
| 1494 |
+
"nll_loss": 0.3642919957637787,
|
| 1495 |
+
"rewards/accuracies": 0.987500011920929,
|
| 1496 |
+
"rewards/chosen": -12.442464828491211,
|
| 1497 |
+
"rewards/margins": 4.127151966094971,
|
| 1498 |
+
"rewards/rejected": -16.569618225097656,
|
| 1499 |
+
"step": 910
|
| 1500 |
+
},
|
| 1501 |
+
{
|
| 1502 |
+
"epoch": 2.1265530193585667,
|
| 1503 |
+
"grad_norm": 8.666687965393066,
|
| 1504 |
+
"learning_rate": 3.2246998284734136e-07,
|
| 1505 |
+
"logits/chosen": -0.3802010118961334,
|
| 1506 |
+
"logits/rejected": -0.37199968099594116,
|
| 1507 |
+
"logps/chosen": -127.65777587890625,
|
| 1508 |
+
"logps/rejected": -168.89389038085938,
|
| 1509 |
+
"loss": 0.4658,
|
| 1510 |
+
"nll_loss": 0.35325732827186584,
|
| 1511 |
+
"rewards/accuracies": 0.96875,
|
| 1512 |
+
"rewards/chosen": -12.765777587890625,
|
| 1513 |
+
"rewards/margins": 4.123614311218262,
|
| 1514 |
+
"rewards/rejected": -16.889392852783203,
|
| 1515 |
+
"step": 920
|
| 1516 |
+
},
|
| 1517 |
+
{
|
| 1518 |
+
"epoch": 2.1496677260907253,
|
| 1519 |
+
"grad_norm": 18.571739196777344,
|
| 1520 |
+
"learning_rate": 3.13893653516295e-07,
|
| 1521 |
+
"logits/chosen": -0.3906049430370331,
|
| 1522 |
+
"logits/rejected": -0.38780367374420166,
|
| 1523 |
+
"logps/chosen": -125.8351058959961,
|
| 1524 |
+
"logps/rejected": -167.4423828125,
|
| 1525 |
+
"loss": 0.4662,
|
| 1526 |
+
"nll_loss": 0.35373836755752563,
|
| 1527 |
+
"rewards/accuracies": 0.9750000238418579,
|
| 1528 |
+
"rewards/chosen": -12.583511352539062,
|
| 1529 |
+
"rewards/margins": 4.160727500915527,
|
| 1530 |
+
"rewards/rejected": -16.744239807128906,
|
| 1531 |
+
"step": 930
|
| 1532 |
+
},
|
| 1533 |
+
{
|
| 1534 |
+
"epoch": 2.1727824328228835,
|
| 1535 |
+
"grad_norm": 13.318151473999023,
|
| 1536 |
+
"learning_rate": 3.053173241852487e-07,
|
| 1537 |
+
"logits/chosen": -0.46741265058517456,
|
| 1538 |
+
"logits/rejected": -0.46371087431907654,
|
| 1539 |
+
"logps/chosen": -136.69558715820312,
|
| 1540 |
+
"logps/rejected": -179.5621795654297,
|
| 1541 |
+
"loss": 0.4453,
|
| 1542 |
+
"nll_loss": 0.35544899106025696,
|
| 1543 |
+
"rewards/accuracies": 0.987500011920929,
|
| 1544 |
+
"rewards/chosen": -13.66955852508545,
|
| 1545 |
+
"rewards/margins": 4.286660194396973,
|
| 1546 |
+
"rewards/rejected": -17.956218719482422,
|
| 1547 |
+
"step": 940
|
| 1548 |
+
},
|
| 1549 |
+
{
|
| 1550 |
+
"epoch": 2.1958971395550417,
|
| 1551 |
+
"grad_norm": 23.31292152404785,
|
| 1552 |
+
"learning_rate": 2.967409948542024e-07,
|
| 1553 |
+
"logits/chosen": -0.5026014447212219,
|
| 1554 |
+
"logits/rejected": -0.4931977391242981,
|
| 1555 |
+
"logps/chosen": -137.78359985351562,
|
| 1556 |
+
"logps/rejected": -178.4962921142578,
|
| 1557 |
+
"loss": 0.458,
|
| 1558 |
+
"nll_loss": 0.3667753040790558,
|
| 1559 |
+
"rewards/accuracies": 0.9781249761581421,
|
| 1560 |
+
"rewards/chosen": -13.778360366821289,
|
| 1561 |
+
"rewards/margins": 4.07127046585083,
|
| 1562 |
+
"rewards/rejected": -17.84963035583496,
|
| 1563 |
+
"step": 950
|
| 1564 |
+
},
|
| 1565 |
+
{
|
| 1566 |
+
"epoch": 2.2190118462872004,
|
| 1567 |
+
"grad_norm": 28.351606369018555,
|
| 1568 |
+
"learning_rate": 2.881646655231561e-07,
|
| 1569 |
+
"logits/chosen": -0.4884619116783142,
|
| 1570 |
+
"logits/rejected": -0.4745559096336365,
|
| 1571 |
+
"logps/chosen": -134.03724670410156,
|
| 1572 |
+
"logps/rejected": -174.16436767578125,
|
| 1573 |
+
"loss": 0.4698,
|
| 1574 |
+
"nll_loss": 0.3588988780975342,
|
| 1575 |
+
"rewards/accuracies": 0.965624988079071,
|
| 1576 |
+
"rewards/chosen": -13.403724670410156,
|
| 1577 |
+
"rewards/margins": 4.012711524963379,
|
| 1578 |
+
"rewards/rejected": -17.41643714904785,
|
| 1579 |
+
"step": 960
|
| 1580 |
+
},
|
| 1581 |
+
{
|
| 1582 |
+
"epoch": 2.2421265530193586,
|
| 1583 |
+
"grad_norm": 15.93598747253418,
|
| 1584 |
+
"learning_rate": 2.7958833619210975e-07,
|
| 1585 |
+
"logits/chosen": -0.5274711847305298,
|
| 1586 |
+
"logits/rejected": -0.5002039074897766,
|
| 1587 |
+
"logps/chosen": -132.9987335205078,
|
| 1588 |
+
"logps/rejected": -171.1938018798828,
|
| 1589 |
+
"loss": 0.4804,
|
| 1590 |
+
"nll_loss": 0.3535648286342621,
|
| 1591 |
+
"rewards/accuracies": 0.9624999761581421,
|
| 1592 |
+
"rewards/chosen": -13.299871444702148,
|
| 1593 |
+
"rewards/margins": 3.8195080757141113,
|
| 1594 |
+
"rewards/rejected": -17.119380950927734,
|
| 1595 |
+
"step": 970
|
| 1596 |
+
},
|
| 1597 |
+
{
|
| 1598 |
+
"epoch": 2.265241259751517,
|
| 1599 |
+
"grad_norm": 13.009766578674316,
|
| 1600 |
+
"learning_rate": 2.710120068610635e-07,
|
| 1601 |
+
"logits/chosen": -0.40121275186538696,
|
| 1602 |
+
"logits/rejected": -0.4012083411216736,
|
| 1603 |
+
"logps/chosen": -134.23101806640625,
|
| 1604 |
+
"logps/rejected": -174.37557983398438,
|
| 1605 |
+
"loss": 0.445,
|
| 1606 |
+
"nll_loss": 0.3591146767139435,
|
| 1607 |
+
"rewards/accuracies": 0.96875,
|
| 1608 |
+
"rewards/chosen": -13.423103332519531,
|
| 1609 |
+
"rewards/margins": 4.014455795288086,
|
| 1610 |
+
"rewards/rejected": -17.437557220458984,
|
| 1611 |
+
"step": 980
|
| 1612 |
+
},
|
| 1613 |
+
{
|
| 1614 |
+
"epoch": 2.2883559664836755,
|
| 1615 |
+
"grad_norm": 10.573274612426758,
|
| 1616 |
+
"learning_rate": 2.6243567753001716e-07,
|
| 1617 |
+
"logits/chosen": -0.3767433762550354,
|
| 1618 |
+
"logits/rejected": -0.3655189275741577,
|
| 1619 |
+
"logps/chosen": -126.58245849609375,
|
| 1620 |
+
"logps/rejected": -166.48191833496094,
|
| 1621 |
+
"loss": 0.4537,
|
| 1622 |
+
"nll_loss": 0.3470018208026886,
|
| 1623 |
+
"rewards/accuracies": 0.971875011920929,
|
| 1624 |
+
"rewards/chosen": -12.658246994018555,
|
| 1625 |
+
"rewards/margins": 3.989945650100708,
|
| 1626 |
+
"rewards/rejected": -16.648191452026367,
|
| 1627 |
+
"step": 990
|
| 1628 |
+
},
|
| 1629 |
+
{
|
| 1630 |
+
"epoch": 2.3114706732158337,
|
| 1631 |
+
"grad_norm": 13.547653198242188,
|
| 1632 |
+
"learning_rate": 2.5385934819897083e-07,
|
| 1633 |
+
"logits/chosen": -0.36574894189834595,
|
| 1634 |
+
"logits/rejected": -0.371765673160553,
|
| 1635 |
+
"logps/chosen": -132.08352661132812,
|
| 1636 |
+
"logps/rejected": -174.651611328125,
|
| 1637 |
+
"loss": 0.4627,
|
| 1638 |
+
"nll_loss": 0.35074400901794434,
|
| 1639 |
+
"rewards/accuracies": 0.981249988079071,
|
| 1640 |
+
"rewards/chosen": -13.208353042602539,
|
| 1641 |
+
"rewards/margins": 4.256808280944824,
|
| 1642 |
+
"rewards/rejected": -17.465160369873047,
|
| 1643 |
+
"step": 1000
|
| 1644 |
+
},
|
| 1645 |
+
{
|
| 1646 |
+
"epoch": 2.334585379947992,
|
| 1647 |
+
"grad_norm": 13.845785140991211,
|
| 1648 |
+
"learning_rate": 2.452830188679245e-07,
|
| 1649 |
+
"logits/chosen": -0.3355964124202728,
|
| 1650 |
+
"logits/rejected": -0.3307664096355438,
|
| 1651 |
+
"logps/chosen": -128.27835083007812,
|
| 1652 |
+
"logps/rejected": -171.4265899658203,
|
| 1653 |
+
"loss": 0.4469,
|
| 1654 |
+
"nll_loss": 0.3528256416320801,
|
| 1655 |
+
"rewards/accuracies": 0.971875011920929,
|
| 1656 |
+
"rewards/chosen": -12.827836990356445,
|
| 1657 |
+
"rewards/margins": 4.314822673797607,
|
| 1658 |
+
"rewards/rejected": -17.142658233642578,
|
| 1659 |
+
"step": 1010
|
| 1660 |
+
},
|
| 1661 |
+
{
|
| 1662 |
+
"epoch": 2.35770008668015,
|
| 1663 |
+
"grad_norm": 14.853008270263672,
|
| 1664 |
+
"learning_rate": 2.367066895368782e-07,
|
| 1665 |
+
"logits/chosen": -0.4388372302055359,
|
| 1666 |
+
"logits/rejected": -0.44896426796913147,
|
| 1667 |
+
"logps/chosen": -140.34243774414062,
|
| 1668 |
+
"logps/rejected": -182.12892150878906,
|
| 1669 |
+
"loss": 0.4646,
|
| 1670 |
+
"nll_loss": 0.3558862805366516,
|
| 1671 |
+
"rewards/accuracies": 0.9781249761581421,
|
| 1672 |
+
"rewards/chosen": -14.0342435836792,
|
| 1673 |
+
"rewards/margins": 4.178647518157959,
|
| 1674 |
+
"rewards/rejected": -18.212890625,
|
| 1675 |
+
"step": 1020
|
| 1676 |
+
},
|
| 1677 |
+
{
|
| 1678 |
+
"epoch": 2.3808147934123087,
|
| 1679 |
+
"grad_norm": 17.43394660949707,
|
| 1680 |
+
"learning_rate": 2.281303602058319e-07,
|
| 1681 |
+
"logits/chosen": -0.35451555252075195,
|
| 1682 |
+
"logits/rejected": -0.36915498971939087,
|
| 1683 |
+
"logps/chosen": -122.70450592041016,
|
| 1684 |
+
"logps/rejected": -161.46353149414062,
|
| 1685 |
+
"loss": 0.4646,
|
| 1686 |
+
"nll_loss": 0.3501938283443451,
|
| 1687 |
+
"rewards/accuracies": 0.9624999761581421,
|
| 1688 |
+
"rewards/chosen": -12.2704496383667,
|
| 1689 |
+
"rewards/margins": 3.875903606414795,
|
| 1690 |
+
"rewards/rejected": -16.14635467529297,
|
| 1691 |
+
"step": 1030
|
| 1692 |
+
},
|
| 1693 |
+
{
|
| 1694 |
+
"epoch": 2.403929500144467,
|
| 1695 |
+
"grad_norm": 12.251300811767578,
|
| 1696 |
+
"learning_rate": 2.1955403087478557e-07,
|
| 1697 |
+
"logits/chosen": -0.5434596538543701,
|
| 1698 |
+
"logits/rejected": -0.4985465407371521,
|
| 1699 |
+
"logps/chosen": -139.0737762451172,
|
| 1700 |
+
"logps/rejected": -181.25820922851562,
|
| 1701 |
+
"loss": 0.4465,
|
| 1702 |
+
"nll_loss": 0.34608450531959534,
|
| 1703 |
+
"rewards/accuracies": 0.96875,
|
| 1704 |
+
"rewards/chosen": -13.907376289367676,
|
| 1705 |
+
"rewards/margins": 4.218443393707275,
|
| 1706 |
+
"rewards/rejected": -18.12582015991211,
|
| 1707 |
+
"step": 1040
|
| 1708 |
+
},
|
| 1709 |
+
{
|
| 1710 |
+
"epoch": 2.427044206876625,
|
| 1711 |
+
"grad_norm": 16.99835205078125,
|
| 1712 |
+
"learning_rate": 2.1097770154373927e-07,
|
| 1713 |
+
"logits/chosen": -0.4769843518733978,
|
| 1714 |
+
"logits/rejected": -0.4647130072116852,
|
| 1715 |
+
"logps/chosen": -141.46798706054688,
|
| 1716 |
+
"logps/rejected": -184.92819213867188,
|
| 1717 |
+
"loss": 0.4619,
|
| 1718 |
+
"nll_loss": 0.34933143854141235,
|
| 1719 |
+
"rewards/accuracies": 0.984375,
|
| 1720 |
+
"rewards/chosen": -14.146799087524414,
|
| 1721 |
+
"rewards/margins": 4.346020698547363,
|
| 1722 |
+
"rewards/rejected": -18.49281883239746,
|
| 1723 |
+
"step": 1050
|
| 1724 |
+
},
|
| 1725 |
+
{
|
| 1726 |
+
"epoch": 2.4501589136087834,
|
| 1727 |
+
"grad_norm": 12.114070892333984,
|
| 1728 |
+
"learning_rate": 2.0240137221269295e-07,
|
| 1729 |
+
"logits/chosen": -0.546917736530304,
|
| 1730 |
+
"logits/rejected": -0.5437088012695312,
|
| 1731 |
+
"logps/chosen": -135.78121948242188,
|
| 1732 |
+
"logps/rejected": -176.3412628173828,
|
| 1733 |
+
"loss": 0.4464,
|
| 1734 |
+
"nll_loss": 0.3510003685951233,
|
| 1735 |
+
"rewards/accuracies": 0.9906250238418579,
|
| 1736 |
+
"rewards/chosen": -13.57812213897705,
|
| 1737 |
+
"rewards/margins": 4.056002616882324,
|
| 1738 |
+
"rewards/rejected": -17.634126663208008,
|
| 1739 |
+
"step": 1060
|
| 1740 |
+
},
|
| 1741 |
+
{
|
| 1742 |
+
"epoch": 2.473273620340942,
|
| 1743 |
+
"grad_norm": 10.714365005493164,
|
| 1744 |
+
"learning_rate": 1.9382504288164663e-07,
|
| 1745 |
+
"logits/chosen": -0.40261998772621155,
|
| 1746 |
+
"logits/rejected": -0.381422221660614,
|
| 1747 |
+
"logps/chosen": -126.05067443847656,
|
| 1748 |
+
"logps/rejected": -166.70254516601562,
|
| 1749 |
+
"loss": 0.5006,
|
| 1750 |
+
"nll_loss": 0.3639281392097473,
|
| 1751 |
+
"rewards/accuracies": 0.971875011920929,
|
| 1752 |
+
"rewards/chosen": -12.605070114135742,
|
| 1753 |
+
"rewards/margins": 4.065185070037842,
|
| 1754 |
+
"rewards/rejected": -16.67025375366211,
|
| 1755 |
+
"step": 1070
|
| 1756 |
+
},
|
| 1757 |
+
{
|
| 1758 |
+
"epoch": 2.4963883270731,
|
| 1759 |
+
"grad_norm": 14.228996276855469,
|
| 1760 |
+
"learning_rate": 1.8524871355060033e-07,
|
| 1761 |
+
"logits/chosen": -0.3770430088043213,
|
| 1762 |
+
"logits/rejected": -0.3675852417945862,
|
| 1763 |
+
"logps/chosen": -127.53874206542969,
|
| 1764 |
+
"logps/rejected": -168.65322875976562,
|
| 1765 |
+
"loss": 0.4691,
|
| 1766 |
+
"nll_loss": 0.35038691759109497,
|
| 1767 |
+
"rewards/accuracies": 0.9750000238418579,
|
| 1768 |
+
"rewards/chosen": -12.753876686096191,
|
| 1769 |
+
"rewards/margins": 4.111447334289551,
|
| 1770 |
+
"rewards/rejected": -16.865324020385742,
|
| 1771 |
+
"step": 1080
|
| 1772 |
+
},
|
| 1773 |
+
{
|
| 1774 |
+
"epoch": 2.5195030338052584,
|
| 1775 |
+
"grad_norm": 14.870238304138184,
|
| 1776 |
+
"learning_rate": 1.76672384219554e-07,
|
| 1777 |
+
"logits/chosen": -0.4105190336704254,
|
| 1778 |
+
"logits/rejected": -0.4109324812889099,
|
| 1779 |
+
"logps/chosen": -122.85371398925781,
|
| 1780 |
+
"logps/rejected": -162.6988525390625,
|
| 1781 |
+
"loss": 0.4539,
|
| 1782 |
+
"nll_loss": 0.34932848811149597,
|
| 1783 |
+
"rewards/accuracies": 0.9906250238418579,
|
| 1784 |
+
"rewards/chosen": -12.285372734069824,
|
| 1785 |
+
"rewards/margins": 3.9845123291015625,
|
| 1786 |
+
"rewards/rejected": -16.269886016845703,
|
| 1787 |
+
"step": 1090
|
| 1788 |
+
},
|
| 1789 |
+
{
|
| 1790 |
+
"epoch": 2.542617740537417,
|
| 1791 |
+
"grad_norm": 18.935836791992188,
|
| 1792 |
+
"learning_rate": 1.6809605488850769e-07,
|
| 1793 |
+
"logits/chosen": -0.3432958424091339,
|
| 1794 |
+
"logits/rejected": -0.3376050591468811,
|
| 1795 |
+
"logps/chosen": -129.51318359375,
|
| 1796 |
+
"logps/rejected": -169.71923828125,
|
| 1797 |
+
"loss": 0.4619,
|
| 1798 |
+
"nll_loss": 0.36302056908607483,
|
| 1799 |
+
"rewards/accuracies": 0.9781249761581421,
|
| 1800 |
+
"rewards/chosen": -12.951319694519043,
|
| 1801 |
+
"rewards/margins": 4.020605564117432,
|
| 1802 |
+
"rewards/rejected": -16.971923828125,
|
| 1803 |
+
"step": 1100
|
| 1804 |
+
},
|
| 1805 |
+
{
|
| 1806 |
+
"epoch": 2.5657324472695753,
|
| 1807 |
+
"grad_norm": 11.919585227966309,
|
| 1808 |
+
"learning_rate": 1.5951972555746142e-07,
|
| 1809 |
+
"logits/chosen": -0.3838345408439636,
|
| 1810 |
+
"logits/rejected": -0.40160542726516724,
|
| 1811 |
+
"logps/chosen": -128.54510498046875,
|
| 1812 |
+
"logps/rejected": -172.9639129638672,
|
| 1813 |
+
"loss": 0.4529,
|
| 1814 |
+
"nll_loss": 0.3525315225124359,
|
| 1815 |
+
"rewards/accuracies": 0.9781249761581421,
|
| 1816 |
+
"rewards/chosen": -12.854510307312012,
|
| 1817 |
+
"rewards/margins": 4.441880226135254,
|
| 1818 |
+
"rewards/rejected": -17.296390533447266,
|
| 1819 |
+
"step": 1110
|
| 1820 |
+
},
|
| 1821 |
+
{
|
| 1822 |
+
"epoch": 2.5888471540017335,
|
| 1823 |
+
"grad_norm": 15.11146354675293,
|
| 1824 |
+
"learning_rate": 1.509433962264151e-07,
|
| 1825 |
+
"logits/chosen": -0.3595576286315918,
|
| 1826 |
+
"logits/rejected": -0.3595134913921356,
|
| 1827 |
+
"logps/chosen": -130.70518493652344,
|
| 1828 |
+
"logps/rejected": -174.2328643798828,
|
| 1829 |
+
"loss": 0.4664,
|
| 1830 |
+
"nll_loss": 0.3525366187095642,
|
| 1831 |
+
"rewards/accuracies": 0.9750000238418579,
|
| 1832 |
+
"rewards/chosen": -13.070520401000977,
|
| 1833 |
+
"rewards/margins": 4.352766990661621,
|
| 1834 |
+
"rewards/rejected": -17.42328453063965,
|
| 1835 |
+
"step": 1120
|
| 1836 |
+
},
|
| 1837 |
+
{
|
| 1838 |
+
"epoch": 2.611961860733892,
|
| 1839 |
+
"grad_norm": 20.21427345275879,
|
| 1840 |
+
"learning_rate": 1.423670668953688e-07,
|
| 1841 |
+
"logits/chosen": -0.3621642291545868,
|
| 1842 |
+
"logits/rejected": -0.3667486906051636,
|
| 1843 |
+
"logps/chosen": -124.94161224365234,
|
| 1844 |
+
"logps/rejected": -166.67266845703125,
|
| 1845 |
+
"loss": 0.449,
|
| 1846 |
+
"nll_loss": 0.34673169255256653,
|
| 1847 |
+
"rewards/accuracies": 0.971875011920929,
|
| 1848 |
+
"rewards/chosen": -12.494161605834961,
|
| 1849 |
+
"rewards/margins": 4.173105239868164,
|
| 1850 |
+
"rewards/rejected": -16.667266845703125,
|
| 1851 |
+
"step": 1130
|
| 1852 |
+
},
|
| 1853 |
+
{
|
| 1854 |
+
"epoch": 2.6350765674660503,
|
| 1855 |
+
"grad_norm": 12.434587478637695,
|
| 1856 |
+
"learning_rate": 1.3379073756432248e-07,
|
| 1857 |
+
"logits/chosen": -0.39929765462875366,
|
| 1858 |
+
"logits/rejected": -0.38376063108444214,
|
| 1859 |
+
"logps/chosen": -134.3514404296875,
|
| 1860 |
+
"logps/rejected": -176.7789764404297,
|
| 1861 |
+
"loss": 0.4449,
|
| 1862 |
+
"nll_loss": 0.34507042169570923,
|
| 1863 |
+
"rewards/accuracies": 0.987500011920929,
|
| 1864 |
+
"rewards/chosen": -13.435147285461426,
|
| 1865 |
+
"rewards/margins": 4.242752552032471,
|
| 1866 |
+
"rewards/rejected": -17.677898406982422,
|
| 1867 |
+
"step": 1140
|
| 1868 |
+
},
|
| 1869 |
+
{
|
| 1870 |
+
"epoch": 2.6581912741982086,
|
| 1871 |
+
"grad_norm": 19.153188705444336,
|
| 1872 |
+
"learning_rate": 1.2521440823327615e-07,
|
| 1873 |
+
"logits/chosen": -0.3830157220363617,
|
| 1874 |
+
"logits/rejected": -0.3753224015235901,
|
| 1875 |
+
"logps/chosen": -138.58120727539062,
|
| 1876 |
+
"logps/rejected": -179.58470153808594,
|
| 1877 |
+
"loss": 0.4505,
|
| 1878 |
+
"nll_loss": 0.3539791703224182,
|
| 1879 |
+
"rewards/accuracies": 0.981249988079071,
|
| 1880 |
+
"rewards/chosen": -13.858120918273926,
|
| 1881 |
+
"rewards/margins": 4.100350379943848,
|
| 1882 |
+
"rewards/rejected": -17.958471298217773,
|
| 1883 |
+
"step": 1150
|
| 1884 |
+
},
|
| 1885 |
+
{
|
| 1886 |
+
"epoch": 2.681305980930367,
|
| 1887 |
+
"grad_norm": 21.320459365844727,
|
| 1888 |
+
"learning_rate": 1.1663807890222984e-07,
|
| 1889 |
+
"logits/chosen": -0.39191287755966187,
|
| 1890 |
+
"logits/rejected": -0.3838447630405426,
|
| 1891 |
+
"logps/chosen": -125.08642578125,
|
| 1892 |
+
"logps/rejected": -166.50216674804688,
|
| 1893 |
+
"loss": 0.4788,
|
| 1894 |
+
"nll_loss": 0.34421923756599426,
|
| 1895 |
+
"rewards/accuracies": 0.965624988079071,
|
| 1896 |
+
"rewards/chosen": -12.508642196655273,
|
| 1897 |
+
"rewards/margins": 4.141573905944824,
|
| 1898 |
+
"rewards/rejected": -16.65021514892578,
|
| 1899 |
+
"step": 1160
|
| 1900 |
+
},
|
| 1901 |
+
{
|
| 1902 |
+
"epoch": 2.7044206876625254,
|
| 1903 |
+
"grad_norm": 14.173407554626465,
|
| 1904 |
+
"learning_rate": 1.0806174957118352e-07,
|
| 1905 |
+
"logits/chosen": -0.3383176028728485,
|
| 1906 |
+
"logits/rejected": -0.32876265048980713,
|
| 1907 |
+
"logps/chosen": -124.4461898803711,
|
| 1908 |
+
"logps/rejected": -165.80368041992188,
|
| 1909 |
+
"loss": 0.4486,
|
| 1910 |
+
"nll_loss": 0.34746435284614563,
|
| 1911 |
+
"rewards/accuracies": 0.984375,
|
| 1912 |
+
"rewards/chosen": -12.444619178771973,
|
| 1913 |
+
"rewards/margins": 4.135747909545898,
|
| 1914 |
+
"rewards/rejected": -16.580368041992188,
|
| 1915 |
+
"step": 1170
|
| 1916 |
+
},
|
| 1917 |
+
{
|
| 1918 |
+
"epoch": 2.7275353943946836,
|
| 1919 |
+
"grad_norm": 17.886056900024414,
|
| 1920 |
+
"learning_rate": 9.948542024013721e-08,
|
| 1921 |
+
"logits/chosen": -0.3328002393245697,
|
| 1922 |
+
"logits/rejected": -0.3495582342147827,
|
| 1923 |
+
"logps/chosen": -131.67807006835938,
|
| 1924 |
+
"logps/rejected": -175.42776489257812,
|
| 1925 |
+
"loss": 0.4633,
|
| 1926 |
+
"nll_loss": 0.3507309556007385,
|
| 1927 |
+
"rewards/accuracies": 0.9750000238418579,
|
| 1928 |
+
"rewards/chosen": -13.167805671691895,
|
| 1929 |
+
"rewards/margins": 4.374970436096191,
|
| 1930 |
+
"rewards/rejected": -17.542776107788086,
|
| 1931 |
+
"step": 1180
|
| 1932 |
+
},
|
| 1933 |
+
{
|
| 1934 |
+
"epoch": 2.750650101126842,
|
| 1935 |
+
"grad_norm": 15.006377220153809,
|
| 1936 |
+
"learning_rate": 9.09090909090909e-08,
|
| 1937 |
+
"logits/chosen": -0.4121033251285553,
|
| 1938 |
+
"logits/rejected": -0.4125909209251404,
|
| 1939 |
+
"logps/chosen": -135.8062286376953,
|
| 1940 |
+
"logps/rejected": -179.68875122070312,
|
| 1941 |
+
"loss": 0.4435,
|
| 1942 |
+
"nll_loss": 0.3616895079612732,
|
| 1943 |
+
"rewards/accuracies": 0.9906250238418579,
|
| 1944 |
+
"rewards/chosen": -13.580622673034668,
|
| 1945 |
+
"rewards/margins": 4.388251304626465,
|
| 1946 |
+
"rewards/rejected": -17.968875885009766,
|
| 1947 |
+
"step": 1190
|
| 1948 |
+
},
|
| 1949 |
+
{
|
| 1950 |
+
"epoch": 2.773764807859,
|
| 1951 |
+
"grad_norm": 15.148149490356445,
|
| 1952 |
+
"learning_rate": 8.23327615780446e-08,
|
| 1953 |
+
"logits/chosen": -0.41161981225013733,
|
| 1954 |
+
"logits/rejected": -0.4115080237388611,
|
| 1955 |
+
"logps/chosen": -131.64926147460938,
|
| 1956 |
+
"logps/rejected": -172.00704956054688,
|
| 1957 |
+
"loss": 0.4567,
|
| 1958 |
+
"nll_loss": 0.3558318614959717,
|
| 1959 |
+
"rewards/accuracies": 0.96875,
|
| 1960 |
+
"rewards/chosen": -13.164926528930664,
|
| 1961 |
+
"rewards/margins": 4.0357770919799805,
|
| 1962 |
+
"rewards/rejected": -17.20070457458496,
|
| 1963 |
+
"step": 1200
|
| 1964 |
+
},
|
| 1965 |
+
{
|
| 1966 |
+
"epoch": 2.7968795145911587,
|
| 1967 |
+
"grad_norm": 14.142349243164062,
|
| 1968 |
+
"learning_rate": 7.375643224699828e-08,
|
| 1969 |
+
"logits/chosen": -0.38675186038017273,
|
| 1970 |
+
"logits/rejected": -0.3759027123451233,
|
| 1971 |
+
"logps/chosen": -119.2894058227539,
|
| 1972 |
+
"logps/rejected": -160.90139770507812,
|
| 1973 |
+
"loss": 0.4422,
|
| 1974 |
+
"nll_loss": 0.3399142920970917,
|
| 1975 |
+
"rewards/accuracies": 0.9750000238418579,
|
| 1976 |
+
"rewards/chosen": -11.92894172668457,
|
| 1977 |
+
"rewards/margins": 4.161198616027832,
|
| 1978 |
+
"rewards/rejected": -16.090139389038086,
|
| 1979 |
+
"step": 1210
|
| 1980 |
+
},
|
| 1981 |
+
{
|
| 1982 |
+
"epoch": 2.819994221323317,
|
| 1983 |
+
"grad_norm": 12.57529354095459,
|
| 1984 |
+
"learning_rate": 6.518010291595197e-08,
|
| 1985 |
+
"logits/chosen": -0.3739423155784607,
|
| 1986 |
+
"logits/rejected": -0.3440755009651184,
|
| 1987 |
+
"logps/chosen": -126.98980712890625,
|
| 1988 |
+
"logps/rejected": -168.39813232421875,
|
| 1989 |
+
"loss": 0.4573,
|
| 1990 |
+
"nll_loss": 0.347236305475235,
|
| 1991 |
+
"rewards/accuracies": 0.965624988079071,
|
| 1992 |
+
"rewards/chosen": -12.698979377746582,
|
| 1993 |
+
"rewards/margins": 4.140833377838135,
|
| 1994 |
+
"rewards/rejected": -16.839813232421875,
|
| 1995 |
+
"step": 1220
|
| 1996 |
+
},
|
| 1997 |
+
{
|
| 1998 |
+
"epoch": 2.843108928055475,
|
| 1999 |
+
"grad_norm": 14.81961727142334,
|
| 2000 |
+
"learning_rate": 5.660377358490566e-08,
|
| 2001 |
+
"logits/chosen": -0.3412761092185974,
|
| 2002 |
+
"logits/rejected": -0.3169342875480652,
|
| 2003 |
+
"logps/chosen": -121.49079895019531,
|
| 2004 |
+
"logps/rejected": -163.16494750976562,
|
| 2005 |
+
"loss": 0.4468,
|
| 2006 |
+
"nll_loss": 0.35947078466415405,
|
| 2007 |
+
"rewards/accuracies": 0.965624988079071,
|
| 2008 |
+
"rewards/chosen": -12.149078369140625,
|
| 2009 |
+
"rewards/margins": 4.167415618896484,
|
| 2010 |
+
"rewards/rejected": -16.31649398803711,
|
| 2011 |
+
"step": 1230
|
| 2012 |
+
},
|
| 2013 |
+
{
|
| 2014 |
+
"epoch": 2.8662236347876338,
|
| 2015 |
+
"grad_norm": 17.703699111938477,
|
| 2016 |
+
"learning_rate": 4.802744425385934e-08,
|
| 2017 |
+
"logits/chosen": -0.3797430396080017,
|
| 2018 |
+
"logits/rejected": -0.3891495168209076,
|
| 2019 |
+
"logps/chosen": -130.4372100830078,
|
| 2020 |
+
"logps/rejected": -171.90744018554688,
|
| 2021 |
+
"loss": 0.4626,
|
| 2022 |
+
"nll_loss": 0.35077929496765137,
|
| 2023 |
+
"rewards/accuracies": 0.971875011920929,
|
| 2024 |
+
"rewards/chosen": -13.043721199035645,
|
| 2025 |
+
"rewards/margins": 4.147024631500244,
|
| 2026 |
+
"rewards/rejected": -17.190744400024414,
|
| 2027 |
+
"step": 1240
|
| 2028 |
+
},
|
| 2029 |
+
{
|
| 2030 |
+
"epoch": 2.889338341519792,
|
| 2031 |
+
"grad_norm": 12.169320106506348,
|
| 2032 |
+
"learning_rate": 3.945111492281304e-08,
|
| 2033 |
+
"logits/chosen": -0.3306048512458801,
|
| 2034 |
+
"logits/rejected": -0.31452488899230957,
|
| 2035 |
+
"logps/chosen": -128.30259704589844,
|
| 2036 |
+
"logps/rejected": -170.43658447265625,
|
| 2037 |
+
"loss": 0.4428,
|
| 2038 |
+
"nll_loss": 0.35800302028656006,
|
| 2039 |
+
"rewards/accuracies": 0.96875,
|
| 2040 |
+
"rewards/chosen": -12.830259323120117,
|
| 2041 |
+
"rewards/margins": 4.213398456573486,
|
| 2042 |
+
"rewards/rejected": -17.043659210205078,
|
| 2043 |
+
"step": 1250
|
| 2044 |
+
},
|
| 2045 |
+
{
|
| 2046 |
+
"epoch": 2.91245304825195,
|
| 2047 |
+
"grad_norm": 11.698453903198242,
|
| 2048 |
+
"learning_rate": 3.087478559176672e-08,
|
| 2049 |
+
"logits/chosen": -0.38229602575302124,
|
| 2050 |
+
"logits/rejected": -0.3816959857940674,
|
| 2051 |
+
"logps/chosen": -131.13912963867188,
|
| 2052 |
+
"logps/rejected": -173.8229522705078,
|
| 2053 |
+
"loss": 0.4627,
|
| 2054 |
+
"nll_loss": 0.3486027419567108,
|
| 2055 |
+
"rewards/accuracies": 0.9781249761581421,
|
| 2056 |
+
"rewards/chosen": -13.113912582397461,
|
| 2057 |
+
"rewards/margins": 4.268382549285889,
|
| 2058 |
+
"rewards/rejected": -17.382295608520508,
|
| 2059 |
+
"step": 1260
|
| 2060 |
+
},
|
| 2061 |
+
{
|
| 2062 |
+
"epoch": 2.935567754984109,
|
| 2063 |
+
"grad_norm": 12.507227897644043,
|
| 2064 |
+
"learning_rate": 2.229845626072041e-08,
|
| 2065 |
+
"logits/chosen": -0.3767063617706299,
|
| 2066 |
+
"logits/rejected": -0.3743255138397217,
|
| 2067 |
+
"logps/chosen": -133.4464569091797,
|
| 2068 |
+
"logps/rejected": -176.45843505859375,
|
| 2069 |
+
"loss": 0.4712,
|
| 2070 |
+
"nll_loss": 0.3417053818702698,
|
| 2071 |
+
"rewards/accuracies": 0.981249988079071,
|
| 2072 |
+
"rewards/chosen": -13.344644546508789,
|
| 2073 |
+
"rewards/margins": 4.301196098327637,
|
| 2074 |
+
"rewards/rejected": -17.64583969116211,
|
| 2075 |
+
"step": 1270
|
| 2076 |
+
},
|
| 2077 |
+
{
|
| 2078 |
+
"epoch": 2.958682461716267,
|
| 2079 |
+
"grad_norm": 17.885986328125,
|
| 2080 |
+
"learning_rate": 1.3722126929674098e-08,
|
| 2081 |
+
"logits/chosen": -0.40772175788879395,
|
| 2082 |
+
"logits/rejected": -0.41167011857032776,
|
| 2083 |
+
"logps/chosen": -131.9138946533203,
|
| 2084 |
+
"logps/rejected": -176.85263061523438,
|
| 2085 |
+
"loss": 0.4666,
|
| 2086 |
+
"nll_loss": 0.3591021001338959,
|
| 2087 |
+
"rewards/accuracies": 0.9781249761581421,
|
| 2088 |
+
"rewards/chosen": -13.191390991210938,
|
| 2089 |
+
"rewards/margins": 4.49387264251709,
|
| 2090 |
+
"rewards/rejected": -17.685260772705078,
|
| 2091 |
+
"step": 1280
|
| 2092 |
+
},
|
| 2093 |
+
{
|
| 2094 |
+
"epoch": 2.9817971684484252,
|
| 2095 |
+
"grad_norm": 12.335119247436523,
|
| 2096 |
+
"learning_rate": 5.145797598627788e-09,
|
| 2097 |
+
"logits/chosen": -0.3768162131309509,
|
| 2098 |
+
"logits/rejected": -0.37075626850128174,
|
| 2099 |
+
"logps/chosen": -136.2458038330078,
|
| 2100 |
+
"logps/rejected": -178.2454071044922,
|
| 2101 |
+
"loss": 0.4426,
|
| 2102 |
+
"nll_loss": 0.3536157011985779,
|
| 2103 |
+
"rewards/accuracies": 0.9906250238418579,
|
| 2104 |
+
"rewards/chosen": -13.624580383300781,
|
| 2105 |
+
"rewards/margins": 4.1999616622924805,
|
| 2106 |
+
"rewards/rejected": -17.824541091918945,
|
| 2107 |
+
"step": 1290
|
| 2108 |
+
},
|
| 2109 |
+
{
|
| 2110 |
+
"epoch": 2.9956659924877203,
|
| 2111 |
+
"eval_logits/chosen": -0.34079235792160034,
|
| 2112 |
+
"eval_logits/rejected": -0.31415677070617676,
|
| 2113 |
+
"eval_logps/chosen": -143.9447784423828,
|
| 2114 |
+
"eval_logps/rejected": -155.83648681640625,
|
| 2115 |
+
"eval_loss": 1.5984355211257935,
|
| 2116 |
+
"eval_nll_loss": 0.3937048017978668,
|
| 2117 |
+
"eval_rewards/accuracies": 0.6304348111152649,
|
| 2118 |
+
"eval_rewards/chosen": -14.394478797912598,
|
| 2119 |
+
"eval_rewards/margins": 1.1891697645187378,
|
| 2120 |
+
"eval_rewards/rejected": -15.583648681640625,
|
| 2121 |
+
"eval_runtime": 73.5734,
|
| 2122 |
+
"eval_samples_per_second": 24.819,
|
| 2123 |
+
"eval_steps_per_second": 1.563,
|
| 2124 |
+
"step": 1296
|
| 2125 |
+
},
|
| 2126 |
+
{
|
| 2127 |
+
"epoch": 2.9956659924877203,
|
| 2128 |
+
"step": 1296,
|
| 2129 |
+
"total_flos": 0.0,
|
| 2130 |
+
"train_loss": 1.025652496903031,
|
| 2131 |
+
"train_runtime": 27913.7357,
|
| 2132 |
+
"train_samples_per_second": 5.951,
|
| 2133 |
+
"train_steps_per_second": 0.046
|
| 2134 |
+
}
|
| 2135 |
+
],
|
| 2136 |
+
"logging_steps": 10,
|
| 2137 |
+
"max_steps": 1296,
|
| 2138 |
+
"num_input_tokens_seen": 0,
|
| 2139 |
+
"num_train_epochs": 3,
|
| 2140 |
+
"save_steps": 500,
|
| 2141 |
+
"stateful_callbacks": {
|
| 2142 |
+
"TrainerControl": {
|
| 2143 |
+
"args": {
|
| 2144 |
+
"should_epoch_stop": false,
|
| 2145 |
+
"should_evaluate": false,
|
| 2146 |
+
"should_log": false,
|
| 2147 |
+
"should_save": true,
|
| 2148 |
+
"should_training_stop": true
|
| 2149 |
+
},
|
| 2150 |
+
"attributes": {}
|
| 2151 |
+
}
|
| 2152 |
+
},
|
| 2153 |
+
"total_flos": 0.0,
|
| 2154 |
+
"train_batch_size": 4,
|
| 2155 |
+
"trial_name": null,
|
| 2156 |
+
"trial_params": null
|
| 2157 |
+
}
|