See axolotl config

axolotl version: 0.4.1

adapter: lora
base_model: Qwen/Qwen2.5-1.5B-Instruct
bf16: true
chat_template: llama3
dataset_prepared_path: null
datasets:
- data_files:
  - 6a67cd14306afa65_train_data.json
  ds_type: json
  format: custom
  path: /workspace/input_data/6a67cd14306afa65_train_data.json
  type:
    field_input: schema
    field_instruction: question
    field_output: cypher
    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/56d3ea50-6039-40d4-9fc0-7daef7b67dda
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: 32
lora_dropout: 0.05
lora_fan_in_fan_out: null
lora_model_dir: null
lora_r: 16
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: 4140
micro_batch_size: 4
mlflow_experiment_name: /tmp/6a67cd14306afa65_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: 0.04
wandb_entity: null
wandb_mode: online
wandb_name: 524a0a62-67eb-4359-bad2-51bcaeeb1570
wandb_project: Gradients-On-Demand
wandb_run: your_name
wandb_runid: 524a0a62-67eb-4359-bad2-51bcaeeb1570
warmup_steps: 10
weight_decay: 0.0
xformers_attention: null

56d3ea50-6039-40d4-9fc0-7daef7b67dda

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

• Loss: 0.1067

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: 2598

Training results

Training Loss	Epoch	Step	Validation Loss
1.531	0.0008	1	1.7833
0.1689	0.0770	100	0.2207
0.1654	0.1540	200	0.1798
0.1234	0.2310	300	0.1621
0.1609	0.3080	400	0.1573
0.1739	0.3850	500	0.1438
0.1142	0.4620	600	0.1419
0.0944	0.5390	700	0.1363
0.1433	0.6160	800	0.1366
0.1062	0.6930	900	0.1293
0.1382	0.7700	1000	0.1255
0.1003	0.8470	1100	0.1234
0.1454	0.9241	1200	0.1188
0.1326	1.0013	1300	0.1181
0.1031	1.0783	1400	0.1161
0.0339	1.1553	1500	0.1159
0.1357	1.2323	1600	0.1157
0.081	1.3093	1700	0.1137
0.1138	1.3863	1800	0.1127
0.0779	1.4633	1900	0.1114
0.0797	1.5403	2000	0.1093
0.0863	1.6173	2100	0.1086
0.059	1.6943	2200	0.1075
0.1495	1.7713	2300	0.1072
0.065	1.8483	2400	0.1069
0.1151	1.9253	2500	0.1067

Framework versions

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