File size: 3,179 Bytes
a8966c5 61b8f7e a8966c5 465e398 a8966c5 465e398 a8966c5 465e398 a8966c5 bce5b0e a8966c5 61b8f7e |
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 61 62 63 64 65 66 67 68 69 70 71 72 73 74 75 76 77 78 79 80 81 82 83 84 85 86 87 88 89 90 91 92 93 94 95 96 97 98 99 100 101 102 103 104 105 106 107 108 109 110 111 112 113 114 115 116 117 118 119 120 121 122 123 124 125 126 127 128 129 130 131 132 133 134 135 136 137 138 139 140 |
---
tags:
- generated_from_trainer
model-index:
- name: out
results: []
datasets:
- teknium/openhermes
---
This is the Instruction Fine Tuned version of [Tiny Llama](https://github.com/jzhang38/TinyLlama) on [@Teknium1's](https://twitter.com/Teknium1) [openhermes](https://huggingface.co/datasets/teknium/openhermes) dataset.
`"The TinyLlama project aims to pretrain a 1.1B Llama model on 3 trillion tokens. With some proper optimization, we can achieve this within a span of "just" 90 days using 16 A100-40G GPUs 🚀🚀. The training has started on 2023-09-01."`
<details><summary>See axolotl config</summary>
axolotl version: `0.3.0`
```yaml
base_model: ./TinyLlama-1.1B-intermediate-step-1431k-3T
model_type: LlamaForCausalLM
tokenizer_type: LlamaTokenizer
is_llama_derived_model: true
load_in_8bit: false
load_in_4bit: false
strict: false
datasets:
- path: ./openhermes
type: alpaca
dataset_prepared_path:
val_set_size: 0.05
output_dir: ./out
sequence_len: 4096
sample_packing: false
adapter:
lora_model_dir:
lora_r:
lora_alpha:
lora_dropout:
lora_target_linear:
lora_fan_in_fan_out:
wandb_project: tinyllama-openhermes
wandb_entity: tensoic
wandb_watch:
wandb_name:
wandb_log_model:
gradient_accumulation_steps: 2
micro_batch_size: 8
num_epochs: 1
optimizer: adamw_bnb_8bit
adam_epsilon: 0.00001
max_grad_norm: 1.0
lr_scheduler: cosine
learning_rate: 0.0002
train_on_inputs: false
group_by_length: false
bf16: false
fp16: true
tf32: false
gradient_checkpointing: true
early_stopping_patience:
resume_from_checkpoint:
local_rank:
logging_steps: 1
xformers_attention: true
flash_attention:
warmup_steps: 100
evals_per_epoch: 4
eval_table_size:
saves_per_epoch: 1
debug:
deepspeed: zero2.json
weight_decay: 0.0
fsdp:
fsdp_config:
special_tokens:
bos_token: "<s>"
eos_token: "</s>"
unk_token: "<unk>"
```
</details><br>
The model achieves the following loss:
- Loss: 1.3647
The loss exploded after a couple hundred steps. As suggested by [winglian](https://x.com/winglian/status/1740776666744700941?s=20), we set the following values in the config file:
```
adam_epsilon: 0.00001
max_grad_norm: 1.0
```
### Training hyperparameters
The following hyperparameters were used during training:
- learning_rate: 0.0002
- train_batch_size: 8
- eval_batch_size: 8
- seed: 42
- distributed_type: multi-GPU
- num_devices: 8
- gradient_accumulation_steps: 2
- total_train_batch_size: 128
- total_eval_batch_size: 64
- optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-05
- lr_scheduler_type: cosine
- lr_scheduler_warmup_steps: 100
- num_epochs: 1
- mixed_precision_training: Native AMP
### Training results
| Training Loss | Epoch | Step | Validation Loss |
|:-------------:|:-----:|:----:|:---------------:|
| 3.0006 | 0.0 | 1 | 1.6838 |
| 0.8195 | 0.25 | 451 | 1.4620 |
| 0.6836 | 0.5 | 902 | 1.4158 |
| 0.6811 | 0.75 | 1353 | 1.3647 |
### Framework versions
- Transformers 4.36.2
- Pytorch 2.0.1+cu117
- Datasets 2.15.0
- Tokenizers 0.15.0 |