See axolotl config

axolotl version: 0.8.0.dev0

base_model: paloalma/Qwen-2.5-72B-Instruct-Pruned
hub_model_id: paloalma/Le_Triomphant_ECE_TW3_V2
trust_remote_code: true
load_in_8bit: false
load_in_4bit: true
datasets:
  - path: paloalma/Reasoning-DeepSeek-R1-Distilled-1.4M-Alpaca-20K
    type: alpaca
val_set_size: 0.1
output_dir: ./outputs/deepseek_finetune
sequence_len: 2048
sample_packing: true
pad_to_sequence_len: true
eval_sample_packing: false  # Changed from true to false
adapter: qlora
lora_r: 32
lora_alpha: 64
lora_dropout: 0.05
lora_target_linear: true
chat_template: alpaca
gradient_accumulation_steps: 4
micro_batch_size: 2
num_epochs: 1
saves_per_epoch: 3
learning_rate: 0.0002
tf32: true
gradient_checkpointing: true
deepspeed: deepspeed_configs/zero3.json

Le_Triomphant_ECE_TW3_V2

This model is a fine-tuned version of paloalma/Qwen-2.5-72B-Instruct-Pruned on the paloalma/Reasoning-DeepSeek-R1-Distilled-1.4M-Alpaca-20K dataset. It achieves the following results on the evaluation set:

• Loss: 0.4271

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
• distributed_type: multi-GPU
• num_devices: 3
• gradient_accumulation_steps: 4
• total_train_batch_size: 24
• total_eval_batch_size: 6
• optimizer: Use OptimizerNames.ADAMW_HF 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: 8
• num_epochs: 1.0

Training results

Training Loss	Epoch	Step	Validation Loss
0.4347	0.9973	281	0.4271

Framework versions

• PEFT 0.14.0
• Transformers 4.49.0
• Pytorch 2.5.1+cu124
• Datasets 3.2.0
• Tokenizers 0.21.0