See axolotl config

axolotl version: 0.4.1

adapter: lora
base_model: Qwen/Qwen2.5-0.5B-Instruct
bf16: true
chat_template: llama3
dataset_prepared_path: null
datasets:
- data_files:
  - 9f7cef554ae67229_train_data.json
  ds_type: json
  format: custom
  path: /workspace/input_data/9f7cef554ae67229_train_data.json
  type:
    field_instruction: question
    field_output: answer
    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/11a99da6-e19a-4dc7-82a1-029204f1b39e
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: 8193
micro_batch_size: 4
mlflow_experiment_name: /tmp/9f7cef554ae67229_train_data.json
model_type: AutoModelForCausalLM
num_epochs: 10
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: 0.023880026745629956
wandb_entity: null
wandb_mode: online
wandb_name: 365fa35d-c7f9-424f-8597-3e48c6b82259
wandb_project: Gradients-On-Demand
wandb_run: your_name
wandb_runid: 365fa35d-c7f9-424f-8597-3e48c6b82259
warmup_steps: 10
weight_decay: 0.0
xformers_attention: null

11a99da6-e19a-4dc7-82a1-029204f1b39e

This model is a fine-tuned version of Qwen/Qwen2.5-0.5B-Instruct on the None dataset. It achieves the following results on the evaluation set:

• Loss: 1.9004

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: 8193

Training results

Training Loss	Epoch	Step	Validation Loss
2.7295	0.0002	1	2.6862
2.1482	0.0157	100	2.2432
2.1636	0.0313	200	2.1992
2.0887	0.0470	300	2.1710
2.2107	0.0626	400	2.1557
2.2463	0.0783	500	2.1387
2.1664	0.0939	600	2.1258
2.0681	0.1096	700	2.1143
1.983	0.1253	800	2.1041
2.1054	0.1409	900	2.0956
2.0935	0.1566	1000	2.0910
2.0639	0.1722	1100	2.0823
2.0159	0.1879	1200	2.0740
2.0348	0.2035	1300	2.0692
2.0211	0.2192	1400	2.0634
2.1148	0.2349	1500	2.0560
2.0104	0.2505	1600	2.0520
2.1911	0.2662	1700	2.0444
2.0186	0.2818	1800	2.0405
2.041	0.2975	1900	2.0358
2.0536	0.3131	2000	2.0346
1.9777	0.3288	2100	2.0268
2.1097	0.3445	2200	2.0237
1.9466	0.3601	2300	2.0198
2.0087	0.3758	2400	2.0161
2.0104	0.3914	2500	2.0109
2.0467	0.4071	2600	2.0068
1.9831	0.4227	2700	2.0044
1.8847	0.4384	2800	1.9994
2.0466	0.4541	2900	1.9954
1.9743	0.4697	3000	1.9910
1.9384	0.4854	3100	1.9869
1.9385	0.5010	3200	1.9833
1.9445	0.5167	3300	1.9803
2.0068	0.5323	3400	1.9769
1.8507	0.5480	3500	1.9755
2.0212	0.5637	3600	1.9710
1.9067	0.5793	3700	1.9669
2.0083	0.5950	3800	1.9650
1.9992	0.6106	3900	1.9604
1.9096	0.6263	4000	1.9567
1.924	0.6419	4100	1.9536
1.9568	0.6576	4200	1.9502
1.9162	0.6733	4300	1.9476
2.0678	0.6889	4400	1.9440
1.9842	0.7046	4500	1.9413
2.0508	0.7202	4600	1.9381
1.9671	0.7359	4700	1.9357
1.9338	0.7515	4800	1.9325
1.9149	0.7672	4900	1.9296
1.8924	0.7829	5000	1.9276
2.0384	0.7985	5100	1.9249
1.9198	0.8142	5200	1.9227
1.9877	0.8298	5300	1.9203
1.9321	0.8455	5400	1.9175
1.8519	0.8611	5500	1.9162
1.9102	0.8768	5600	1.9133
1.976	0.8925	5700	1.9116
1.8236	0.9081	5800	1.9097
1.8497	0.9238	5900	1.9076
1.991	0.9394	6000	1.9060
1.9287	0.9551	6100	1.9043
1.9014	0.9707	6200	1.9025
1.836	0.9864	6300	1.9011
1.7582	1.0021	6400	1.9003
1.8858	1.0178	6500	1.9012
1.8173	1.0334	6600	1.9004

Framework versions

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