See axolotl config

axolotl version: 0.4.1

adapter: lora
base_model: unsloth/Qwen2-0.5B-Instruct
bf16: true
chat_template: llama3
dataset_prepared_path: null
datasets:
- data_files:
  - 19fd35b02e02d35a_train_data.json
  ds_type: json
  format: custom
  path: /workspace/input_data/19fd35b02e02d35a_train_data.json
  type:
    field_input: input
    field_instruction: instruction
    field_output: output
    format: '{instruction} {input}'
    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/dd25756c-6a8d-4ec0-b8a1-b1f456f6a333
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: 8832
micro_batch_size: 4
mlflow_experiment_name: /tmp/19fd35b02e02d35a_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: 0.044897409419476494
wandb_entity: null
wandb_mode: online
wandb_name: b3cd6cf2-8402-4373-a1f6-7aa530c7ed80
wandb_project: Gradients-On-Demand
wandb_run: your_name
wandb_runid: b3cd6cf2-8402-4373-a1f6-7aa530c7ed80
warmup_steps: 10
weight_decay: 0.0
xformers_attention: null

dd25756c-6a8d-4ec0-b8a1-b1f456f6a333

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

• Loss: 1.9745

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: 6648

Training results

Training Loss	Epoch	Step	Validation Loss
2.5935	0.0003	1	2.7137
2.4756	0.0301	100	2.4414
2.417	0.0602	200	2.3848
2.1331	0.0903	300	2.3462
2.0593	0.1203	400	2.3140
2.0063	0.1504	500	2.2910
2.4628	0.1805	600	2.2694
2.2041	0.2106	700	2.2530
2.4011	0.2407	800	2.2392
2.2987	0.2708	900	2.2226
2.199	0.3008	1000	2.2099
2.2245	0.3309	1100	2.1960
2.375	0.3610	1200	2.1850
2.2182	0.3911	1300	2.1771
2.3893	0.4212	1400	2.1658
2.1014	0.4513	1500	2.1578
2.1474	0.4813	1600	2.1484
2.4473	0.5114	1700	2.1396
1.9483	0.5415	1800	2.1326
2.1937	0.5716	1900	2.1209
2.2298	0.6017	2000	2.1139
2.1117	0.6318	2100	2.1069
2.2471	0.6619	2200	2.0990
2.1825	0.6919	2300	2.0947
2.1731	0.7220	2400	2.0892
1.8862	0.7521	2500	2.0825
2.1224	0.7822	2600	2.0744
1.9015	0.8123	2700	2.0710
2.103	0.8424	2800	2.0637
2.0056	0.8724	2900	2.0575
1.8938	0.9025	3000	2.0523
2.1503	0.9326	3100	2.0460
2.2166	0.9627	3200	2.0415
2.1761	0.9928	3300	2.0358
1.9747	1.0229	3400	2.0398
1.6468	1.0529	3500	2.0353
1.7083	1.0830	3600	2.0323
1.9831	1.1131	3700	2.0292
1.8527	1.1432	3800	2.0236
1.9907	1.1733	3900	2.0209
1.9898	1.2034	4000	2.0193
1.9063	1.2335	4100	2.0153
1.674	1.2635	4200	2.0101
1.7583	1.2936	4300	2.0083
2.076	1.3237	4400	2.0045
1.92	1.3538	4500	2.0034
2.0666	1.3839	4600	1.9988
1.8152	1.4140	4700	1.9958
1.6996	1.4440	4800	1.9938
1.7863	1.4741	4900	1.9926
1.9677	1.5042	5000	1.9888
1.9768	1.5343	5100	1.9879
1.7981	1.5644	5200	1.9857
1.7892	1.5945	5300	1.9841
1.8826	1.6245	5400	1.9830
1.8107	1.6546	5500	1.9810
2.01	1.6847	5600	1.9790
1.789	1.7148	5700	1.9787
1.6017	1.7449	5800	1.9773
1.8574	1.7750	5900	1.9767
1.695	1.8051	6000	1.9758
1.8974	1.8351	6100	1.9752
1.7432	1.8652	6200	1.9752
1.7931	1.8953	6300	1.9748
1.9937	1.9254	6400	1.9747
2.2055	1.9555	6500	1.9746
1.8637	1.9856	6600	1.9745

Framework versions

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