See axolotl config

axolotl version: 0.4.1

adapter: lora
base_model: Qwen/Qwen2-1.5B-Instruct
bf16: auto
chat_template: llama3
dataset_prepared_path: null
datasets:
- data_files:
  - d778b6b213e10026_train_data.json
  ds_type: json
  format: custom
  path: /workspace/input_data/d778b6b213e10026_train_data.json
  type:
    field_input: context
    field_instruction: question
    field_output: answer
    format: '{instruction} {input}'
    no_input_format: '{instruction}'
    system_format: '{system}'
    system_prompt: ''
debug: null
deepspeed: null
device: cuda
early_stopping_patience: 1
eval_max_new_tokens: 128
eval_steps: 5
eval_table_size: null
evals_per_epoch: null
flash_attention: false
fp16: null
gradient_accumulation_steps: 4
gradient_checkpointing: true
gradient_clipping: 1.0
group_by_length: true
hub_model_id: infogep/808a6e92-a8a8-4fce-a8ba-0fe1fb0fa0e8
hub_repo: null
hub_strategy: end
hub_token: null
learning_rate: 5.0e-05
load_in_4bit: false
load_in_8bit: false
local_rank: null
logging_steps: 3
lora_alpha: 16
lora_dropout: 0.05
lora_fan_in_fan_out: null
lora_model_dir: null
lora_r: 8
lora_target_linear: true
lr_scheduler: cosine
max_steps: 30
micro_batch_size: 4
mlflow_experiment_name: /tmp/d778b6b213e10026_train_data.json
model_type: AutoModelForCausalLM
num_epochs: 1
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: 10
sequence_len: 1024
strict: false
tf32: false
tokenizer_type: AutoTokenizer
train_on_inputs: true
trust_remote_code: true
val_set_size: 0.05
wandb_entity: null
wandb_mode: online
wandb_name: ac3fbae1-18ab-431e-bf64-c4cfae3e3134
wandb_project: Gradients-On-Demand
wandb_run: your_name
wandb_runid: ac3fbae1-18ab-431e-bf64-c4cfae3e3134
warmup_steps: 5
weight_decay: 0.0
xformers_attention: true

808a6e92-a8a8-4fce-a8ba-0fe1fb0fa0e8

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

• Loss: 1.1104

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: 5e-05
• train_batch_size: 4
• eval_batch_size: 4
• seed: 42
• gradient_accumulation_steps: 4
• total_train_batch_size: 16
• 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: 5
• training_steps: 30

Training results

Training Loss	Epoch	Step	Validation Loss
No log	0.0001	1	1.7403
0.2011	0.0006	5	1.7286
1.2771	0.0011	10	1.5541
1.3929	0.0017	15	1.2981
1.2744	0.0023	20	1.1668
1.4062	0.0028	25	1.1190
1.2905	0.0034	30	1.1104

Framework versions

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