End of training

README.md

@@ -44,7 +44,7 @@ More information needed
 
 # Resource Usage Comparison
 
-- VRAM Use: 8.0719 GB
+- VRAM Use: 7.4136 GB
 
 # Distillation (Teacher -> Student) Architecture Difference:
 
@@ -75,7 +75,7 @@ More information needed
 <br/>
 
 # Train Dataset
-Trained on 226,123,153 tokens from the [wikimedia/wikipedia](https://huggingface.co/datasets/wikimedia/wikipedia) dataset.
+Trained on 226,081,419 tokens from the [wikimedia/wikipedia](https://huggingface.co/datasets/wikimedia/wikipedia) dataset.
 
 - Num Samples: `396,000`
 - Subset: `20231101.en`
@@ -85,7 +85,7 @@ Trained on 226,123,153 tokens from the [wikimedia/wikipedia](https://huggingface
 # Training Objective
 
 ```
-DistillationObjective(logits_loss_component=LossComponent(label=logits, weight=1, loss_fn=kl), attn_loss_component=LossComponent(label=attn, weight=5, loss_fn=raw_mse, layer_mapper=layer-2, norm=layernorm_teacher_only, projector=orthogonal))
+DistillationObjective(logits_loss_component=LossComponent(label=logits, weight=1, loss_fn=kl), attn_loss_component=LossComponent(label=attn, weight=5, loss_fn=raw_mse, layer_mapper=layer-2, norm=layernorm_teacher_only, projector=mlp))
 ```
 
 # Hyperparameters
@@ -95,15 +95,15 @@ The following hyperparameters were used during training:
 <summary>Expand</summary>
 
 - learning_rate: `0.0002`
-- train_batch_size: `8`
+- train_batch_size: `4`
 - eval_batch_size: `8`
 - seed: `42`
 - optimizer: `Adam with betas=(0.9,0.999) and epsilon=1e-08`
 - lr_scheduler_type: `polynomial`
 - num_epochs: `1.0`
-- distillation_objective: `DistillationObjective(logits_loss_component=LossComponent(label=logits, weight=1, loss_fn=kl), attn_loss_component=LossComponent(label=attn, weight=5, loss_fn=raw_mse, layer_mapper=layer-2, norm=layernorm_teacher_only, projector=orthogonal))`
+- distillation_objective: `DistillationObjective(logits_loss_component=LossComponent(label=logits, weight=1, loss_fn=kl), attn_loss_component=LossComponent(label=attn, weight=5, loss_fn=raw_mse, layer_mapper=layer-2, norm=layernorm_teacher_only, projector=mlp))`
 - train_embeddings: `True`
-- lr_scheduler: `<torch.optim.lr_scheduler.LambdaLR object at 0x7f930c64e9b0>`
+- lr_scheduler: `<torch.optim.lr_scheduler.LambdaLR object at 0x7f35ec2f7b20>`
 - student_model_name_or_path: `None`
 - student_config_name_or_path: `distilbert/distilgpt2`
 - student_model_config: `None`
@@ -134,5 +134,5 @@ The following hyperparameters were used during training:
 # Framework Versions
 - Distily 0.4.1
 - Transformers 4.44.2
-- Pytorch 2.3.0
-- Datasets 2.21.0
+- Pytorch 2.4.0+cu121
+- Datasets 2.18.0

logs/attn_norm=layernorm_teacher_only, attn_projector=mlp, attn_weight=5, learning_rate=0.0002, per_device_train_batch_size=4, warmup_ratio=0/events.out.tfevents.1725127984.a7e428977e35

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:3aa1e6cf20b78488e28c9ec4f921233f8667794588eded084e657351deb43f39
+size 47486066

logs/attn_norm=layernorm_teacher_only, attn_projector=mlp, attn_weight=5, learning_rate=0.0002, per_device_train_batch_size=4, warmup_ratio=0/events.out.tfevents.1725137271.a7e428977e35

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:24070be04cfabbf68dea1d28965f78e746354204a389d5bceeb72815ac626586
+size 529

model.safetensors

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:d5d7d2ba009d5d2d75800a059124ededa189094647f9d8de4422fce358e618eb
+oid sha256:b9b3bb893ff8ee4666cd233538d884bc5ccd5b1d3ec0d206f66f7886cbcc2604
 size 163832792

training_args.bin

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:ca1184841742ca5a1521e047576aca845588ce26266cf49b16becc7a01700fb7
-size 5624
+oid sha256:0d4f75e6f5aceb8b2a9ec32350f6923c51af093499fcf34b437a6b3422ed7671
+size 5560