See axolotl config

axolotl version: 0.4.1

adapter: qlora
auto_resume_from_checkpoints: true
base_model: Qwen/Qwen2-7B-Instruct
bf16: auto
chat_template: llama3
dataset_prepared_path: null
dataset_processes: 6
datasets:
- data_files:
  - 8128ac1cac761e50_train_data.json
  ds_type: json
  format: custom
  path: /workspace/input_data/8128ac1cac761e50_train_data.json
  type:
    field_instruction: prompt
    field_output: completion
    format: '{instruction}'
    no_input_format: '{instruction}'
    system_format: '{system}'
    system_prompt: ''
debug: null
deepspeed: null
early_stopping_patience: 5
eval_max_new_tokens: 128
eval_steps: 200
eval_table_size: null
evals_per_epoch: null
flash_attention: true
fp16: false
fsdp: null
fsdp_config: null
gradient_accumulation_steps: 4
gradient_checkpointing: true
group_by_length: false
hub_model_id: error577/130f3f97-04a9-4e26-a9f6-b9fa4b6d816c
hub_repo: null
hub_strategy: checkpoint
hub_token: null
learning_rate: 0.0002
load_in_4bit: true
load_in_8bit: false
local_rank: null
logging_steps: 1
lora_alpha: 64
lora_dropout: 0.1
lora_fan_in_fan_out: null
lora_model_dir: null
lora_r: 32
lora_target_linear: true
lr_scheduler: cosine
max_grad_norm: 1.0
max_steps: null
micro_batch_size: 2
mlflow_experiment_name: /tmp/8128ac1cac761e50_train_data.json
model_type: AutoModelForCausalLM
num_epochs: 3
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: 200
sequence_len: 256
strict: false
tf32: false
tokenizer_type: AutoTokenizer
train_on_inputs: false
trust_remote_code: true
val_set_size: 0.005
wandb_entity: null
wandb_mode: online
wandb_name: eb41b3cb-964b-41fb-b1ba-824d594e7af4
wandb_project: Gradients-On-Demand
wandb_run: your_name
wandb_runid: eb41b3cb-964b-41fb-b1ba-824d594e7af4
warmup_steps: 30
weight_decay: 0.0
xformers_attention: null

130f3f97-04a9-4e26-a9f6-b9fa4b6d816c

This model is a fine-tuned version of Qwen/Qwen2-7B-Instruct on the None dataset. It achieves the following results on the evaluation set:

• Loss: 2.0677

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: 2
• eval_batch_size: 2
• seed: 42
• gradient_accumulation_steps: 4
• total_train_batch_size: 8
• 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: 30
• num_epochs: 3

Training results

Training Loss	Epoch	Step	Validation Loss
1.8207	0.0001	1	2.5730
0.6464	0.0244	200	0.4704
1.0167	0.0488	400	0.4498
0.4289	0.0732	600	0.4360
0.4053	0.0975	800	0.4169
0.1816	0.1219	1000	0.4446
0.7136	0.1463	1200	0.4180
0.9619	0.1707	1400	0.4166
0.1941	0.1951	1600	0.4163
0.0499	0.2195	1800	0.4229
0.2176	0.2439	2000	0.4192
0.9006	0.2682	2200	0.4098
10.7892	0.2926	2400	7.6201
3.1323	0.3170	2600	3.2112
5.2014	0.3414	2800	2.6625
3.3664	0.3658	3000	2.4255
3.5745	0.3902	3200	2.0677

Framework versions

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