Training in progress, step 49500

README.md

@@ -44,7 +44,7 @@ More information needed
 
 # Resource Usage Comparison
 
-- VRAM Use: 8.0694 GB
+- VRAM Use: 8.0719 GB
 
 # Distillation (Teacher -> Student) Architecture Difference:
 
@@ -75,7 +75,7 @@ More information needed
 <br/>
 
 # Train Dataset
-Trained on 226,096,614 tokens from the [wikimedia/wikipedia](https://huggingface.co/datasets/wikimedia/wikipedia) dataset.
+Trained on 226,120,971 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 226,096,614 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=rmsnorm_teacher_only, projector=mlp))
+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, projector=orthogonal))
 ```
 
 # Hyperparameters
@@ -101,9 +101,9 @@ The following hyperparameters were used during training:
 - 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=rmsnorm_teacher_only, projector=mlp))`
+- 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, projector=orthogonal))`
 - train_embeddings: `True`
-- lr_scheduler: `<torch.optim.lr_scheduler.LambdaLR object at 0x7f1d54320940>`
+- lr_scheduler: `<torch.optim.lr_scheduler.LambdaLR object at 0x7f69141039a0>`
 - 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 2.4.0+cu121
+- Pytorch 2.3.0
 - Datasets 2.21.0

logs/attn_norm=instancenorm, attn_projector=orthogonal, attn_weight=5, learning_rate=0.0002, per_device_train_batch_size=8, warmup_ratio=0/events.out.tfevents.1725162765.849724f928d2

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:773a0812c51636ddf755cbc84c24d083c3f370e58d83774a9d1cd77b1d7234f6
+size 23671670

logs/attn_norm=rmsnorm, attn_projector=orthogonal, attn_weight=5, learning_rate=0.0002, per_device_train_batch_size=8, warmup_ratio=0/events.out.tfevents.1725162642.849724f928d2

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:5a4be4d2068250adb38638220a3cc35625faff37702acdede2c3dce8d70b8de7
+size 5780

logs/attn_norm=rmsnorm_teacher_only, attn_projector=orthogonal, attn_weight=5, learning_rate=0.0002, per_device_train_batch_size=8, warmup_ratio=0/events.out.tfevents.1725162703.849724f928d2

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:4a0a5f9bdb136c2b2d9e567699d5d614a56f0ca9e76c9fbc2244c1eb1b708ff3
+size 5806

model.safetensors

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:e009bcbe94cbcf085220b58b1988f2bbfd58d04cee070e863cd2aeea75e89324
+oid sha256:37928efaf6e1955069debf9065abe3795dbaaf09ac1735d23f28a942a3008baf
 size 163832792

training_args.bin

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:19cbc7edb308033267ca96c6b16559e82d71b1933a1aa68262c696efa5d2a325
+oid sha256:d6d2c2667d6e25d0055e7408133a5d519bff47baef4365bdc3c0e5227047da28
 size 5560