base_model: HuggingFaceTB/SmolLM2-135M
model_type: LlamaForCausalLM
tokenizer_type: GPT2Tokenizer

load_in_4bit: true
load_in_8bit: false
strict: false

save_safetensors: true
flash_attention: true
auto_resume_from_checkpoints: true
save_steps: 100
learning_rate: 5e-4

num_epochs: 2


hub_model_id: minpeter/LoRA-SmolLM2-135M-ChatML-Instruct


micro_batch_size: 8
gradient_accumulation_steps: 4

dataset_processes: 1000
chat_template: chatml

datasets:
  - path: vicgalle/alpaca-gpt4
    type: alpaca
  # - path: shibing624/sharegpt_gpt4
  #   type: chat_template
  #   field_messages: conversations
  #   message_field_role: from
  #   message_field_content: value
  #   roles_to_train: ["assistant", "gpt"]
  #   fraction: 0.1

adapter: qlora
lora_r: 16
lora_alpha: 32
lora_dropout: 0.1
lora_target_linear: true
lora_modules_to_save:
  - lm_head
  - embed_tokens


special_tokens:
  bos_token: <|begin_of_text|>
  eos_token: <|end_of_text|> 
  pad_token: <|custom_pad|>
  unk_token: <|custom_unk|>


optimizer: adamw_torch_fused
lr_scheduler: cosine

wandb_project: "axolotl"
wandb_entity: "kasfiekfs-e"

LoRA-SmolLM2-135M-ChatML-Instruct

This model is a fine-tuned version of HuggingFaceTB/SmolLM2-135M on the vicgalle/alpaca-gpt4 dataset.

Model description

More information needed

Intended uses & limitations

More information needed

Training and evaluation data

More information needed

Training procedure

Training hyperparameters

The following hyperparameters were used during training:

learning_rate: 0.0005
train_batch_size: 8
eval_batch_size: 8
seed: 42
gradient_accumulation_steps: 4
total_train_batch_size: 32
optimizer: Use OptimizerNames.ADAMW_TORCH_FUSED with betas=(0.9,0.999) and epsilon=1e-08 and optimizer_args=No additional optimizer arguments
lr_scheduler_type: cosine
lr_scheduler_warmup_steps: 97
num_epochs: 2

Training results

Framework versions

PEFT 0.14.0
Transformers 4.47.1
Pytorch 2.5.1+cu124
Datasets 3.2.0
Tokenizers 0.21.0

minpeter/QLoRA-SmolLM2-135M-Instruct-chatml

LoRA-SmolLM2-135M-ChatML-Instruct

Model description

Intended uses & limitations

Training and evaluation data

Training procedure

Training hyperparameters

Training results

Framework versions

minpeter/
QLoRA-SmolLM2-135M-Instruct-chatml