dywoo/axolotl-Qwen2.5-14B-n2253-LoRA

See axolotl config

axolotl version: 0.5.0

base_model: Qwen/Qwen2.5-14B-Instruct
trust_remote_code: true

load_in_8bit: false
load_in_4bit: false
strict: false

datasets:
  - path: main_dataset_v1.json
    type: alpaca

special_tokens:
  bos_token:
  eos_token: "<|im_end|>"
  pad_token: "<|endoftext|>"

dataset_prepared_path:
val_set_size: 0.05
output_dir: ./outputs/lora-out

sequence_len: 1024
sample_packing: false
pad_to_sequence_len: true

adapter: lora
lora_model_dir:
lora_r: 8
lora_alpha: 16
lora_dropout: 0.05
lora_target_linear: true
lora_fan_in_fan_out:
lora_target_modules:
  - gate_proj
  - down_proj
  - up_proj
  - q_proj
  - v_proj
  - k_proj
  - o_proj

wandb_project: dywoo_axolotl
wandb_entity: dywoo
wandb_watch:
wandb_run_id: 
wandb_log_model:

gradient_accumulation_steps: 2
micro_batch_size: 2
num_epochs: 3
optimizer: paged_adamw_32bit
lr_scheduler: cosine
learning_rate: 0.00005
train_on_inputs:
group_by_length: false
bf16: true
fp16: false
tf32: false
gradient_checkpointing: true
early_stopping_patience:
logging_steps: 100
xformers_attention:
flash_attention: true
warmup_ratio: 0.01
eval_steps: 100
save_steps: 100
save_total_limit: 2
eval_sample_packing:
debug:
deepspeed:
weight_decay: 0.01
fsdp:
fsdp_config:

outputs/lora-out

This model is a fine-tuned version of Qwen/Qwen2.5-14B-Instruct on the None dataset. It achieves the following results on the evaluation set:

• Loss: 0.0749

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: 5e-05
• train_batch_size: 2
• eval_batch_size: 2
• seed: 42
• gradient_accumulation_steps: 2
• total_train_batch_size: 4
• optimizer: Use OptimizerNames.PAGED_ADAMW 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: 16
• num_epochs: 3

Training results

Training Loss	Epoch	Step	Validation Loss
No log	0.0019	1	0.3101
0.1179	0.1869	100	0.0830
0.0312	0.3738	200	0.0780
0.0276	0.5607	300	0.0743
0.0256	0.7477	400	0.0692
0.0222	0.9346	500	0.0705
0.0199	1.1215	600	0.0686
0.0174	1.3084	700	0.0695
0.015	1.4953	800	0.0702
0.0158	1.6822	900	0.0721
0.0147	1.8692	1000	0.0706
0.0139	2.0561	1100	0.0701
0.0097	2.2430	1200	0.0739
0.0099	2.4299	1300	0.0745
0.0097	2.6168	1400	0.0745
0.0107	2.8037	1500	0.0746
0.0093	2.9907	1600	0.0749

Framework versions

• PEFT 0.13.2
• Transformers 4.46.1
• Pytorch 2.3.1+cu121
• Datasets 3.0.1
• Tokenizers 0.20.3