HamdanXI/w2v2_uclass_clipped_10_seconds_fb_labeled

README.md

@@ -1,6 +1,6 @@
 ---
-license: apache-2.0
-base_model: facebook/wav2vec2-base-960h
+license: mit
+base_model: facebook/w2v-bert-2.0
 tags:
 - generated_from_trainer
 metrics:
@@ -15,18 +15,10 @@ should probably proofread and complete it, then remove this comment. -->
 
 # results
 
-This model is a fine-tuned version of [facebook/wav2vec2-base-960h](https://huggingface.co/facebook/wav2vec2-base-960h) on an unknown dataset.
+This model is a fine-tuned version of [facebook/w2v-bert-2.0](https://huggingface.co/facebook/w2v-bert-2.0) on an unknown dataset.
 It achieves the following results on the evaluation set:
-- Loss: 1.5527
-- Accuracy: 0.5206
-- F1 Score Class 0: 0.0
-- F1 Score Class 1: 0.0
-- F1 Score Class 2: 0.0
-- F1 Score Class 3: 0.0
-- F1 Score Class 4: 0.0
-- F1 Score Class 5: 0.0
-- F1 Score Class 6: 0.6847
-- F1 Score Class 7: 0.0
+- Loss: 1.6035
+- Accuracy: 0.5153
 
 ## Model description
 
@@ -49,24 +41,17 @@ The following hyperparameters were used during training:
 - train_batch_size: 16
 - eval_batch_size: 8
 - seed: 42
+- gradient_accumulation_steps: 2
+- total_train_batch_size: 32
 - optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
 - lr_scheduler_type: linear
-- num_epochs: 10
+- num_epochs: 3
 
 ### Training results
 
-| Training Loss | Epoch | Step | Validation Loss | Accuracy | F1 Score Class 0 | F1 Score Class 1 | F1 Score Class 2 | F1 Score Class 3 | F1 Score Class 4 | F1 Score Class 5 | F1 Score Class 6 | F1 Score Class 7 |
-|:-------------:|:-----:|:----:|:---------------:|:--------:|:----------------:|:----------------:|:----------------:|:----------------:|:----------------:|:----------------:|:----------------:|:----------------:|
-| 1.624         | 1.0   | 533  | 1.5562          | 0.5206   | 0.0              | 0.0              | 0.0              | 0.0              | 0.0              | 0.0              | 0.6847           | 0.0              |
-| 1.6758        | 2.0   | 1066 | 1.5635          | 0.5206   | 0.0              | 0.0              | 0.0              | 0.0              | 0.0              | 0.0              | 0.6847           | 0.0              |
-| 1.5986        | 3.0   | 1599 | 1.5559          | 0.5206   | 0.0              | 0.0              | 0.0              | 0.0              | 0.0              | 0.0              | 0.6847           | 0.0              |
-| 1.722         | 4.0   | 2132 | 1.5539          | 0.5206   | 0.0              | 0.0              | 0.0              | 0.0              | 0.0              | 0.0              | 0.6847           | 0.0              |
-| 1.7561        | 5.0   | 2665 | 1.5558          | 0.5206   | 0.0              | 0.0              | 0.0              | 0.0              | 0.0              | 0.0              | 0.6847           | 0.0              |
-| 1.724         | 6.0   | 3198 | 1.5543          | 0.5206   | 0.0              | 0.0              | 0.0              | 0.0              | 0.0              | 0.0              | 0.6847           | 0.0              |
-| 1.485         | 7.0   | 3731 | 1.5534          | 0.5206   | 0.0              | 0.0              | 0.0              | 0.0              | 0.0              | 0.0              | 0.6847           | 0.0              |
-| 1.6086        | 8.0   | 4264 | 1.5524          | 0.5206   | 0.0              | 0.0              | 0.0              | 0.0              | 0.0              | 0.0              | 0.6847           | 0.0              |
-| 1.4925        | 9.0   | 4797 | 1.5530          | 0.5206   | 0.0              | 0.0              | 0.0              | 0.0              | 0.0              | 0.0              | 0.6847           | 0.0              |
-| 1.611         | 10.0  | 5330 | 1.5527          | 0.5206   | 0.0              | 0.0              | 0.0              | 0.0              | 0.0              | 0.0              | 0.6847           | 0.0              |
+| Training Loss | Epoch  | Step | Validation Loss | Accuracy |
+|:-------------:|:------:|:----:|:---------------:|:--------:|
+| 1.6178        | 1.8762 | 500  | 1.6035          | 0.5153   |
 
 
 ### Framework versions

config.json

@@ -1,20 +1,23 @@
 {
-  "_name_or_path": "facebook/wav2vec2-base-960h",
-  "activation_dropout": 0.1,
+  "_name_or_path": "facebook/w2v-bert-2.0",
+  "activation_dropout": 0.0,
+  "adapter_act": "relu",
   "adapter_attn_dim": null,
   "adapter_kernel_size": 3,
   "adapter_stride": 2,
   "add_adapter": false,
-  "apply_spec_augment": true,
+  "apply_spec_augment": false,
   "architectures": [
     "Wav2Vec2ForSequenceClassification"
   ],
-  "attention_dropout": 0.1,
+  "attention_dropout": 0.0,
   "bos_token_id": 1,
-  "classifier_proj_size": 256,
-  "codevector_dim": 256,
+  "classifier_proj_size": 768,
+  "codevector_dim": 768,
+  "conformer_conv_dropout": 0.1,
   "contrastive_logits_temperature": 0.1,
   "conv_bias": false,
+  "conv_depthwise_kernel_size": 31,
   "conv_dim": [
     512,
     512,
@@ -42,22 +45,20 @@
     2,
     2
   ],
-  "ctc_loss_reduction": "sum",
+  "ctc_loss_reduction": "mean",
   "ctc_zero_infinity": false,
   "diversity_loss_weight": 0.1,
   "do_stable_layer_norm": false,
   "eos_token_id": 2,
   "feat_extract_activation": "gelu",
-  "feat_extract_dropout": 0.0,
   "feat_extract_norm": "group",
-  "feat_proj_dropout": 0.1,
+  "feat_proj_dropout": 0.0,
   "feat_quantizer_dropout": 0.0,
+  "feature_projection_input_dim": 160,
   "final_dropout": 0.1,
-  "gradient_checkpointing": false,
-  "hidden_act": "gelu",
-  "hidden_dropout": 0.1,
-  "hidden_dropout_prob": 0.1,
-  "hidden_size": 768,
+  "hidden_act": "swish",
+  "hidden_dropout": 0.0,
+  "hidden_size": 1024,
   "id2label": {
     "0": "LABEL_0",
     "1": "LABEL_1",
@@ -70,7 +71,7 @@
     "8": "LABEL_8"
   },
   "initializer_range": 0.02,
-  "intermediate_size": 3072,
+  "intermediate_size": 4096,
   "label2id": {
     "LABEL_0": 0,
     "LABEL_1": 1,
@@ -83,26 +84,31 @@
     "LABEL_8": 8
   },
   "layer_norm_eps": 1e-05,
-  "layerdrop": 0.1,
+  "layerdrop": 0.0,
+  "left_max_position_embeddings": 64,
   "mask_feature_length": 10,
   "mask_feature_min_masks": 0,
   "mask_feature_prob": 0.0,
   "mask_time_length": 10,
   "mask_time_min_masks": 2,
-  "mask_time_prob": 0.05,
+  "mask_time_prob": 0.0,
+  "max_source_positions": 5000,
   "model_type": "wav2vec2",
-  "num_adapter_layers": 3,
-  "num_attention_heads": 12,
+  "num_adapter_layers": 1,
+  "num_attention_heads": 16,
   "num_codevector_groups": 2,
   "num_codevectors_per_group": 320,
   "num_conv_pos_embedding_groups": 16,
   "num_conv_pos_embeddings": 128,
   "num_feat_extract_layers": 7,
-  "num_hidden_layers": 12,
+  "num_hidden_layers": 24,
   "num_negatives": 100,
-  "output_hidden_size": 768,
-  "pad_token_id": 0,
-  "proj_codevector_dim": 256,
+  "output_hidden_size": 1024,
+  "pad_token_id": 35,
+  "position_embeddings_type": "relative_key",
+  "proj_codevector_dim": 768,
+  "right_max_position_embeddings": 8,
+  "rotary_embedding_base": 10000,
   "tdnn_dilation": [
     1,
     2,
@@ -126,7 +132,8 @@
   ],
   "torch_dtype": "float32",
   "transformers_version": "4.40.1",
+  "use_intermediate_ffn_before_adapter": false,
   "use_weighted_layer_sum": false,
-  "vocab_size": 32,
+  "vocab_size": 38,
   "xvector_output_dim": 512
 }

model.safetensors

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:69223aa1be85c1a0727093742650a369b3fb89d1d372a34bbb17993c0a676b80
-size 378309564
+oid sha256:c159da6c2d11f10f26a32899dd2df5a7c3442071bdfe57fd2b164b3a63caaad6
+size 1264938628

runs/May02_19-13-42_66f4a5d44daa/events.out.tfevents.1714677223.66f4a5d44daa.5906.0

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:0aac6040cbc11b8c363cb8818b1865f5328134681c9b79fede2e5326d0c6e96c
+size 7670

training_args.bin

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:80f29a588cfee9ddf0c86aeeddc61a626fa3f2b90b1fd7182a97da1cebc319f0
-size 4920
+oid sha256:4040a63e96aa775e39a9be00e3638449b75ba13b4591b9397ffbbc6ec482a8bc
+size 4984