Hugging Face's logo

distily
/
distily_smollm_dataset_sweep

TensorBoard
Safetensors
Distily
llama
Generated from Trainer
Model card Files Metrics Community
distily_smollm_dataset_sweep / README.md
lapp0's picture
lapp0
End of training
86f60e4 verified
preview code
|
raw
history blame
7.72 kB
metadata
base_model: HuggingFaceTB/SmolLM-135M
datasets:
  - HuggingFaceFW/fineweb
library_name: Distily
license: creativeml-openrail-m
tags:
  - generated_from_trainer
  - Distily
base_model_relation: finetune
model-index:
  - name: distily_smollm_dataset_sweep
    results: []

Summary

Distilled with Distily library using teacher model HuggingFaceTB/SmolLM-135M on dataset HuggingFaceFW/fineweb.

Model Architecture:

  • Architecture: LlamaForCausalLM
  • Total Parameters: 81,413,568
  • Data Type (dtype): torch.float32
  • Model Size: 0.30 GB
Student Model Details 
LlamaForCausalLM(
  (model): LlamaModel(
    (embed_tokens): Embedding(49152, 576)
    (layers): ModuleList(
      (0-14): 15 x LlamaDecoderLayer(
        (self_attn): LlamaSdpaAttention(
          (q_proj): Linear(in_features=576, out_features=576, bias=False)
          (k_proj): Linear(in_features=576, out_features=192, bias=False)
          (v_proj): Linear(in_features=576, out_features=192, bias=False)
          (o_proj): Linear(in_features=576, out_features=576, bias=False)
          (rotary_emb): LlamaRotaryEmbedding()
        )
        (mlp): LigerSwiGLUMLP(
          (gate_proj): Linear(in_features=576, out_features=1536, bias=False)
          (up_proj): Linear(in_features=576, out_features=1536, bias=False)
          (down_proj): Linear(in_features=1536, out_features=576, bias=False)
        )
        (input_layernorm): LigerRMSNorm((576,), eps=1e-05, offset=0.0)
        (post_attention_layernorm): LigerRMSNorm((576,), eps=1e-05, offset=0.0)
      )
    )
    (norm): LigerRMSNorm((576,), eps=1e-05, offset=0.0)
    (rotary_emb): LlamaRotaryEmbedding()
  )
  (lm_head): Linear(in_features=576, out_features=49152, bias=False)
)

Benchmark Metrics Comparison

Metric distily_smollm_dataset_sweep/logs/dataset_max_seq_length=1024, dataset_sample_size=1000000, dataset_subset=20231101.en, dataset_uri=wikimedia_wikipedia, per_device_train_batch_size=8 distily_smollm_dataset_sweep/logs/dataset_max_seq_length=1024, dataset_sample_size=1000000, dataset_subset=None, dataset_uri=distily_filtered_redpajama_en, per_device_train_batch_size=8 distily_smollm_dataset_sweep/logs/dataset_max_seq_length=1024, dataset_sample_size=1000000, dataset_subset=sample-10BT, dataset_uri=HuggingFaceFW_fineweb, per_device_train_batch_size=8 distily_smollm_dataset_sweep/logs/dataset_max_seq_length=1024, dataset_sample_size=1000000, dataset_subset=sample-10BT, dataset_uri=HuggingFaceFW_fineweb-edu, per_device_train_batch_size=8 logs/teacher
tinyArc.acc_norm,none 0.303 0.295 0.26 0.302 0.37
tinyGSM8k.exact_match,flexible-extract 0.029 0.03 0.006 0.025 0.006
tinyGSM8k.exact_match,strict-match 0.006 0.006 0.006 0.006 0.006
tinyHellaswag.acc_norm,none 0.341 0.281 0.3 0.327 0.452
tinyMMLU.acc_norm,none 0.276 0.281 0.286 0.31 0.341
tinyTruthfulQA.acc,none 0.463 0.447 0.419 0.423 0.38
tinyWinogrande.acc_norm,none 0.466 0.436 0.492 0.46 0.509

Resource Usage

  • Max Train VRAM Use: 13.1269 GB
  • Available VRAM: 23.4329 GB
  • GPUs:
    • 1x NVIDIA GeForce RTX 4090
  • CPUs: 64
  • CPU Memory: 251.7299 GB
  • CPU Memory Bandwidth: 1600 GB/s

Distillation (Teacher -> Student) Architecture Difference:

  • Architecture: LlamaForCausalLM -> LlamaForCausalLM
  • Total Parameters: 134,515,008 -> 81,413,568
  • Data Type (dtype): torch.float32 -> torch.float32
  • Model Size: 0.25 GB -> 0.30 GB
Module Diff Details 
--- teacher model modules
+++ student model modules
@@ -2,7 +2,7 @@
   (model): LlamaModel(
     (embed_tokens): Embedding(49152, 576)
     (layers): ModuleList(
-      (0-29): 30 x LlamaDecoderLayer(
+      (0-14): 15 x LlamaDecoderLayer(
         (self_attn): LlamaSdpaAttention(
           (q_proj): Linear(in_features=576, out_features=576, bias=False)
           (k_proj): Linear(in_features=576, out_features=192, bias=False)
@@ -10,17 +10,16 @@
           (o_proj): Linear(in_features=576, out_features=576, bias=False)
           (rotary_emb): LlamaRotaryEmbedding()
         )
-        (mlp): LlamaMLP(
+        (mlp): LigerSwiGLUMLP(
           (gate_proj): Linear(in_features=576, out_features=1536, bias=False)
           (up_proj): Linear(in_features=576, out_features=1536, bias=False)
           (down_proj): Linear(in_features=1536, out_features=576, bias=False)
-          (act_fn): SiLU()
         )
-        (input_layernorm): LlamaRMSNorm((576,), eps=1e-05)
-        (post_attention_layernorm): LlamaRMSNorm((576,), eps=1e-05)
+        (input_layernorm): LigerRMSNorm((576,), eps=1e-05, offset=0.0)
+        (post_attention_layernorm): LigerRMSNorm((576,), eps=1e-05, offset=0.0)
       )
     )
-    (norm): LlamaRMSNorm((576,), eps=1e-05)
+    (norm): LigerRMSNorm((576,), eps=1e-05, offset=0.0)
     (rotary_emb): LlamaRotaryEmbedding()
   )
   (lm_head): Linear(in_features=576, out_features=49152, bias=False)

Train Dataset

Trained on 501,164,413 tokens from the HuggingFaceFW/fineweb dataset.

  • Num Samples: 998,000
  • Subset: sample-10BT
  • Split: train

Training Objective 

DistillationObjective(
    logits_loss_component=LossComponent(
        weight=1,
        loss_fn='kl'
    ),
    hs_loss_component=LossComponent(
        weight=0
    ),
    attn_loss_component=LossComponent(
        weight=0
    )
)

Hyperparameters

The following hyperparameters were used during training:

Expand
  • learning_rate: 0.0001
  • train_batch_size: 8
  • eval_batch_size: 4
  • seed: 42
  • optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
  • lr_scheduler_type: polynomial
  • lr_scheduler_warmup_ratio: 0.1
  • num_epochs: 1.0
  • distillation_objective: DistillationObjective( logits_loss_component=LossComponent( weight=1, loss_fn='kl' ), hs_loss_component=LossComponent( weight=0 ), attn_loss_component=LossComponent( weight=0 ) )
  • lr_scheduler: <torch.optim.lr_scheduler.LambdaLR object at 0x7205cc5db070>
  • student_model_name_or_path: None
  • student_config_name_or_path: None
  • student_model_config: {'num_hidden_layers': 15}
  • reinitialize_weights: None
  • copy_teacher_modules: [('lm_head', False)]
  • student_model_as_bitnet: False
  • student_use_liger_kernel: True
  • teacher_model_name_or_path: HuggingFaceTB/SmolLM-135M
  • teacher_load_in_8bit: False
  • teacher_load_in_4bit: False
  • dataset_uri: HuggingFaceFW/fineweb
  • dataset_subset: sample-10BT
  • dataset_split: train
  • dataset_column_name: text
  • dataset_sample_size: 1000000
  • dataset_max_seq_length: 1024
  • dataset_test_size: 0.002
  • dataset_shuffle: False
  • dataset_shuffle_seed: 42
  • dataset_trust_remote_code: False
  • gradient_accumulation_steps: 1
  • weight_decay: 0.0
  • max_grad_norm: 1.0
  • warmup_ratio: 0.1
  • warmup_steps: 0
  • gradient_checkpointing: True

Framework Versions

  • Distily 0.5.0
  • Transformers 4.45.0.dev0
  • Pytorch 2.5.0.dev20240910+cu121
  • Datasets 2.21.0