NLP702-bert-large-uncased_peft-distillation_hs768-nh32-nl12

README.md

@@ -15,8 +15,8 @@ should probably proofread and complete it, then remove this comment. -->
 
 This model is a fine-tuned version of [](https://huggingface.co/) on an unknown dataset.
 It achieves the following results on the evaluation set:
-- Loss: 0.4827
-- Accuracy: 0.8386
+- Loss: 0.5089
+- Accuracy: 0.8399
 
 ## Model description
 
@@ -50,10 +50,10 @@ The following hyperparameters were used during training:
 
 | Training Loss | Epoch | Step | Validation Loss | Accuracy |
 |:-------------:|:-----:|:----:|:---------------:|:--------:|
-| 1.8213        | 1.39  | 500  | 0.7941          | 0.7605   |
-| 0.5261        | 2.78  | 1000 | 0.5432          | 0.8249   |
-| 0.2471        | 4.17  | 1500 | 0.4678          | 0.8377   |
-| 0.1381        | 5.56  | 2000 | 0.4495          | 0.8515   |
+| 1.9429        | 1.39  | 500  | 0.9548          | 0.7118   |
+| 0.6755        | 2.78  | 1000 | 0.6468          | 0.8023   |
+| 0.3549        | 4.17  | 1500 | 0.5087          | 0.8411   |
+| 0.1878        | 5.56  | 2000 | 0.4805          | 0.8441   |
 
 
 ### Framework versions

best/config.json

@@ -136,8 +136,8 @@
   "layer_norm_eps": 1e-12,
   "max_position_embeddings": 512,
   "model_type": "bert",
-  "num_attention_heads": 16,
-  "num_hidden_layers": 8,
+  "num_attention_heads": 32,
+  "num_hidden_layers": 12,
   "pad_token_id": 0,
   "position_embedding_type": "absolute",
   "torch_dtype": "float32",

best/model.safetensors

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:5a3ecea4e085f266ba9db26e9ca222babbfb45dd0585fedf6d9b0be19169d726
-size 324723536
+oid sha256:539a36891c29872ca3da1633309b0258c3384693228439c9c1bcd36bb29dac89
+size 438137056

config.json

@@ -136,8 +136,8 @@
   "layer_norm_eps": 1e-12,
   "max_position_embeddings": 512,
   "model_type": "bert",
-  "num_attention_heads": 16,
-  "num_hidden_layers": 8,
+  "num_attention_heads": 32,
+  "num_hidden_layers": 12,
   "pad_token_id": 0,
   "position_embedding_type": "absolute",
   "torch_dtype": "float32",

model.safetensors

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:5a3ecea4e085f266ba9db26e9ca222babbfb45dd0585fedf6d9b0be19169d726
-size 324723536
+oid sha256:539a36891c29872ca3da1633309b0258c3384693228439c9c1bcd36bb29dac89
+size 438137056