base_model: NousResearch/Meta-Llama-3-8B
# optionally might have model_type or tokenizer_type
model_type: LlamaForCausalLM
tokenizer_type: AutoTokenizer
# Automatically upload checkpoint and final model to HF
# hub_model_id: username/custom_model_name

load_in_8bit: true
load_in_4bit: false

datasets:
  - path: cfierro/alpaca-en2fr
    type: alpaca
dataset_prepared_path: /workspace/axolotl-datasets/llama3-en2fr_alpaca
val_set_size: 0.05
output_dir: /workspace/axolotl-outputs/llama3-en2fr_alpaca-lora

sequence_len: 4096
sample_packing: true
eval_sample_packing: false
pad_to_sequence_len: true

adapter: lora
lora_model_dir:
lora_r: 32
lora_alpha: 16
lora_dropout: 0.05
lora_target_linear: true
lora_modules_to_save:
  - embed_tokens
  - lm_head
merge_lora: true

wandb_project: weight-diff-ft
wandb_entity: cfierro
wandb_watch: all
wandb_name: llama3-en2fr_alpaca-lora
wandb_log_model: "false"
hub_model_id: cfierro/llama3-en2fr_alpaca-lora

gradient_accumulation_steps: 4
micro_batch_size: 2
max_steps: 1000  # 8k examples at most 
optimizer: adamw_bnb_8bit
lr_scheduler: cosine
learning_rate: 0.0002

bf16: auto
tf32: false

gradient_checkpointing: true
resume_from_checkpoint:
logging_steps: 1
flash_attention: true

warmup_steps: 10
early_stopping_patience: 2
eval_steps: 50  # bs=4*2 -> eval every 8*50=400 examples  
save_steps: 50  # needed for config validation
save_total_limit: 1
load_best_model_at_end: true
weight_decay: 0.0
special_tokens:
   pad_token: <|end_of_text|>

llama3-en2fr_alpaca-lora

This model is a fine-tuned version of NousResearch/Meta-Llama-3-8B on the cfierro/alpaca-en2fr dataset. It achieves the following results on the evaluation set:

Loss: 0.9983

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: 0.0002
train_batch_size: 2
eval_batch_size: 2
seed: 42
gradient_accumulation_steps: 4
total_train_batch_size: 8
optimizer: Use OptimizerNames.ADAMW_BNB 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: 10
training_steps: 1000

Training results

Training Loss	Epoch	Step	Validation Loss
No log	0	0	1.4635
1.0737	0.2729	50	1.0379
1.0289	0.5457	100	1.0071
1.0187	0.8186	150	0.9880
0.8044	1.0873	200	1.0000
0.8264	1.3602	250	0.9983

Framework versions

PEFT 0.15.2
Transformers 4.53.1
Pytorch 2.7.0+cu128
Datasets 3.6.0
Tokenizers 0.21.2

coastalcph/llama3-en2fr_alpaca-lora

llama3-en2fr_alpaca-lora

Model description

Intended uses & limitations

Training and evaluation data

Training procedure

Training hyperparameters

Training results

Framework versions

coastalcph/
llama3-en2fr_alpaca-lora