See axolotl config

axolotl version: 0.4.1

adapter: lora
base_model: unsloth/Qwen2.5-Coder-1.5B-Instruct
bf16: true
chat_template: llama3
dataset_prepared_path: null
datasets:
- data_files:
  - bdfa47154f25279a_train_data.json
  ds_type: json
  format: custom
  path: /workspace/input_data/bdfa47154f25279a_train_data.json
  type:
    field_instruction: instruction
    field_output: response_8b_instruct
    format: '{instruction}'
    no_input_format: '{instruction}'
    system_format: '{system}'
    system_prompt: ''
debug: null
device_map:
  ? ''
  : 0,1,2,3,4,5,6,7
early_stopping_patience: 2
eval_max_new_tokens: 128
eval_steps: 100
eval_table_size: null
flash_attention: true
gradient_accumulation_steps: 8
gradient_checkpointing: true
group_by_length: false
hub_model_id: Alphatao/81c5b88d-2e4d-4e98-999b-c4ad193eac00
hub_repo: null
hub_strategy: null
hub_token: null
learning_rate: 0.0002
load_best_model_at_end: true
load_in_4bit: false
load_in_8bit: false
local_rank: null
logging_steps: 1
lora_alpha: 128
lora_dropout: 0.1
lora_fan_in_fan_out: null
lora_model_dir: null
lora_r: 64
lora_target_linear: true
lora_target_modules:
- q_proj
- k_proj
- v_proj
- o_proj
lr_scheduler: cosine
max_grad_norm: 1.0
max_steps: 2520
micro_batch_size: 4
mlflow_experiment_name: /tmp/bdfa47154f25279a_train_data.json
model_type: AutoModelForCausalLM
num_epochs: 2
optimizer: adamw_bnb_8bit
output_dir: miner_id_24
pad_to_sequence_len: true
resume_from_checkpoint: null
s2_attention: null
sample_packing: false
save_steps: 100
sequence_len: 1024
strict: false
tf32: true
tokenizer_type: AutoTokenizer
train_on_inputs: false
trust_remote_code: true
val_set_size: 0.05
wandb_entity: null
wandb_mode: online
wandb_name: 610b76a0-719b-4424-a47d-093cf3d53330
wandb_project: Gradients-On-Demand
wandb_run: your_name
wandb_runid: 610b76a0-719b-4424-a47d-093cf3d53330
warmup_steps: 10
weight_decay: 0.0
xformers_attention: null

81c5b88d-2e4d-4e98-999b-c4ad193eac00

This model is a fine-tuned version of unsloth/Qwen2.5-Coder-1.5B-Instruct on the None dataset. It achieves the following results on the evaluation set:

• Loss: 0.6281

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: 4
• eval_batch_size: 4
• seed: 42
• gradient_accumulation_steps: 8
• total_train_batch_size: 32
• 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: 2520

Training results

Training Loss	Epoch	Step	Validation Loss
0.9532	0.0004	1	0.9137
0.7282	0.0414	100	0.7254
0.7217	0.0827	200	0.7107
0.7187	0.1241	300	0.7015
0.6576	0.1655	400	0.6941
0.6551	0.2068	500	0.6868
0.7109	0.2482	600	0.6823
0.7676	0.2895	700	0.6768
0.7061	0.3309	800	0.6723
0.7188	0.3723	900	0.6674
0.6433	0.4136	1000	0.6634
0.6793	0.4550	1100	0.6586
0.7055	0.4964	1200	0.6546
0.6812	0.5377	1300	0.6505
0.6886	0.5791	1400	0.6473
0.6083	0.6204	1500	0.6434
0.6099	0.6618	1600	0.6399
0.7233	0.7032	1700	0.6373
0.6276	0.7445	1800	0.6348
0.6827	0.7859	1900	0.6327
0.6579	0.8273	2000	0.6310
0.6595	0.8686	2100	0.6297
0.6665	0.9100	2200	0.6289
0.5852	0.9513	2300	0.6283
0.5689	0.9927	2400	0.6281
0.4826	1.0342	2500	0.6281

Framework versions

• PEFT 0.13.2
• Transformers 4.46.0
• Pytorch 2.5.0+cu124
• Datasets 3.0.1
• Tokenizers 0.20.1