Training in progress, step 61875
Browse files- README.md +9 -9
- config.json +1 -1
- generation_config.json +1 -1
- logs/attn_layer_mapper=layer-2, attn_loss_fn=raw_mse, attn_norm=layernorm, attn_projector=orthogonal, max_grad_norm=100/completed.flag +0 -0
- logs/attn_layer_mapper=layer-2, attn_loss_fn=raw_mse, attn_norm=layernorm, attn_projector=orthogonal/events.out.tfevents.1724945710.e3f806ea38c9 +3 -0
- model.safetensors +1 -1
- training_args.bin +2 -2
README.md
CHANGED
@@ -46,7 +46,7 @@ More information needed
|
|
46 |
|
47 |
# Resource Usage Comparison
|
48 |
|
49 |
-
- VRAM Use: 7.
|
50 |
|
51 |
# Distillation (Teacher -> Student) Architecture Difference:
|
52 |
|
@@ -66,7 +66,7 @@ More information needed
|
|
66 |
<br/>
|
67 |
|
68 |
# Train Dataset
|
69 |
-
Trained on 145,
|
70 |
|
71 |
- Num Samples: `247,500`
|
72 |
- Subset: `20231101.en`
|
@@ -76,7 +76,7 @@ Trained on 145,744,973 tokens from the [wikimedia/wikipedia](https://huggingface
|
|
76 |
# Training Objective
|
77 |
|
78 |
```
|
79 |
-
DistillationObjective(logits_loss_component=LossComponent(label=logits, weight=1, loss_fn=kl), attn_loss_component=LossComponent(label=attn, weight=25.0, loss_fn=
|
80 |
```
|
81 |
|
82 |
# Hyperparameters
|
@@ -93,9 +93,9 @@ The following hyperparameters were used during training:
|
|
93 |
- lr_scheduler_type: `cosine_with_min_lr`
|
94 |
- lr_scheduler_warmup_ratio: `0.5`
|
95 |
- num_epochs: `1.0`
|
96 |
-
- distillation_objective: `DistillationObjective(logits_loss_component=LossComponent(label=logits, weight=1, loss_fn=kl), attn_loss_component=LossComponent(label=attn, weight=25.0, loss_fn=
|
97 |
- train_embeddings: `True`
|
98 |
-
- lr_scheduler: `<torch.optim.lr_scheduler.LambdaLR object at
|
99 |
- student_model_name_or_path: `None`
|
100 |
- student_config_name_or_path: `None`
|
101 |
- student_model_config: `None`
|
@@ -114,7 +114,7 @@ The following hyperparameters were used during training:
|
|
114 |
- dataset_test_size: `0.01`
|
115 |
- gradient_accumulation_steps: `1`
|
116 |
- weight_decay: `0.0`
|
117 |
-
- max_grad_norm: `
|
118 |
- warmup_ratio: `0.5`
|
119 |
- warmup_steps: `0`
|
120 |
- gradient_checkpointing: `True`
|
@@ -124,7 +124,7 @@ The following hyperparameters were used during training:
|
|
124 |
|
125 |
|
126 |
# Framework Versions
|
127 |
-
- Distily 0.4.
|
128 |
-
- Transformers 4.44.
|
129 |
-
- Pytorch 2.
|
130 |
- Datasets 2.21.0
|
|
|
46 |
|
47 |
# Resource Usage Comparison
|
48 |
|
49 |
+
- VRAM Use: 7.7859 GB
|
50 |
|
51 |
# Distillation (Teacher -> Student) Architecture Difference:
|
52 |
|
|
|
66 |
<br/>
|
67 |
|
68 |
# Train Dataset
|
69 |
+
Trained on 145,716,666 tokens from the [wikimedia/wikipedia](https://huggingface.co/datasets/wikimedia/wikipedia) dataset.
|
70 |
|
71 |
- Num Samples: `247,500`
|
72 |
- Subset: `20231101.en`
|
|
|
76 |
# Training Objective
|
77 |
|
78 |
```
|
79 |
+
DistillationObjective(logits_loss_component=LossComponent(label=logits, weight=1, loss_fn=kl), attn_loss_component=LossComponent(label=attn, weight=25.0, loss_fn=raw_mse, layer_mapper=layer-2, norm=layernorm, projector=orthogonal))
|
80 |
```
|
81 |
|
82 |
# Hyperparameters
|
|
|
93 |
- lr_scheduler_type: `cosine_with_min_lr`
|
94 |
- lr_scheduler_warmup_ratio: `0.5`
|
95 |
- num_epochs: `1.0`
|
96 |
+
- distillation_objective: `DistillationObjective(logits_loss_component=LossComponent(label=logits, weight=1, loss_fn=kl), attn_loss_component=LossComponent(label=attn, weight=25.0, loss_fn=raw_mse, layer_mapper=layer-2, norm=layernorm, projector=orthogonal))`
|
97 |
- train_embeddings: `True`
|
98 |
+
- lr_scheduler: `<torch.optim.lr_scheduler.LambdaLR object at 0x7fe9296249d0>`
|
99 |
- student_model_name_or_path: `None`
|
100 |
- student_config_name_or_path: `None`
|
101 |
- student_model_config: `None`
|
|
|
114 |
- dataset_test_size: `0.01`
|
115 |
- gradient_accumulation_steps: `1`
|
116 |
- weight_decay: `0.0`
|
117 |
+
- max_grad_norm: `100`
|
118 |
- warmup_ratio: `0.5`
|
119 |
- warmup_steps: `0`
|
120 |
- gradient_checkpointing: `True`
|
|
|
124 |
|
125 |
|
126 |
# Framework Versions
|
127 |
+
- Distily 0.4.1
|
128 |
+
- Transformers 4.44.1
|
129 |
+
- Pytorch 2.4.0+cu121
|
130 |
- Datasets 2.21.0
|
config.json
CHANGED
@@ -33,7 +33,7 @@
|
|
33 |
}
|
34 |
},
|
35 |
"torch_dtype": "bfloat16",
|
36 |
-
"transformers_version": "4.44.
|
37 |
"use_cache": true,
|
38 |
"vocab_size": 50257
|
39 |
}
|
|
|
33 |
}
|
34 |
},
|
35 |
"torch_dtype": "bfloat16",
|
36 |
+
"transformers_version": "4.44.1",
|
37 |
"use_cache": true,
|
38 |
"vocab_size": 50257
|
39 |
}
|
generation_config.json
CHANGED
@@ -2,5 +2,5 @@
|
|
2 |
"_from_model_config": true,
|
3 |
"bos_token_id": 50256,
|
4 |
"eos_token_id": 50256,
|
5 |
-
"transformers_version": "4.44.
|
6 |
}
|
|
|
2 |
"_from_model_config": true,
|
3 |
"bos_token_id": 50256,
|
4 |
"eos_token_id": 50256,
|
5 |
+
"transformers_version": "4.44.1"
|
6 |
}
|
logs/attn_layer_mapper=layer-2, attn_loss_fn=raw_mse, attn_norm=layernorm, attn_projector=orthogonal, max_grad_norm=100/completed.flag
ADDED
File without changes
|
logs/attn_layer_mapper=layer-2, attn_loss_fn=raw_mse, attn_norm=layernorm, attn_projector=orthogonal/events.out.tfevents.1724945710.e3f806ea38c9
ADDED
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
1 |
+
version https://git-lfs.github.com/spec/v1
|
2 |
+
oid sha256:e9815948cd4ae167016fa6649d5bc499a6049759526ddbe28f6c63da9a680262
|
3 |
+
size 29625373
|
model.safetensors
CHANGED
@@ -1,3 +1,3 @@
|
|
1 |
version https://git-lfs.github.com/spec/v1
|
2 |
-
oid sha256:
|
3 |
size 248894656
|
|
|
1 |
version https://git-lfs.github.com/spec/v1
|
2 |
+
oid sha256:c8a93d751b1062a819141d33f1f7d6da767b7f02e00f5b3b4914011c1d593afd
|
3 |
size 248894656
|
training_args.bin
CHANGED
@@ -1,3 +1,3 @@
|
|
1 |
version https://git-lfs.github.com/spec/v1
|
2 |
-
oid sha256:
|
3 |
-
size
|
|
|
1 |
version https://git-lfs.github.com/spec/v1
|
2 |
+
oid sha256:4dfbb6d5769cc0f7ba8c61bf1fa23a88db4ffc25e6603c5d68e7604eebfc17db
|
3 |
+
size 5560
|