See axolotl config

axolotl version: 0.4.1

adapter: qlora
base_model: Vikhrmodels/Vikhr-7B-instruct_0.4
bf16: auto
chat_template: llama3
dataloader_num_workers: 6
dataset_prepared_path: null
datasets:
- data_files:
  - 578f14578f1cbc11_train_data.json
  ds_type: json
  format: custom
  path: /workspace/input_data/578f14578f1cbc11_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
deepspeed: null
early_stopping:
  metric: eval_loss
  mode: min
  patience: 3
eval_max_new_tokens: 128
eval_steps: 20
eval_table_size: null
evals_per_epoch: null
flash_attention: true
fp16: null
fsdp: null
fsdp_config: null
gradient_accumulation_steps: 16
gradient_checkpointing: true
group_by_length: true
hub_model_id: error577/0b970ab6-d29b-4eed-bcbf-c4495275f91d
hub_repo: null
hub_strategy: checkpoint
hub_token: null
learning_rate: 0.0003
load_in_4bit: true
load_in_8bit: false
local_rank: null
logging_steps: 1
lora_alpha: 64
lora_dropout: 0.3
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: 300
micro_batch_size: 1
mlflow_experiment_name: /tmp/578f14578f1cbc11_train_data.json
model_type: AutoModelForCausalLM
num_epochs: 50
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: 20
sequence_len: 512
strict: false
tf32: false
tokenizer_type: AutoTokenizer
train_on_inputs: false
trust_remote_code: true
val_set_size: 0.02
wandb_entity: null
wandb_mode: online
wandb_name: dd53892c-9697-4476-ab11-459a7e767cd2
wandb_project: Gradients-On-Demand
wandb_run: your_name
wandb_runid: dd53892c-9697-4476-ab11-459a7e767cd2
warmup_steps: 10
weight_decay: 0.01
xformers_attention: null

0b970ab6-d29b-4eed-bcbf-c4495275f91d

This model is a fine-tuned version of Vikhrmodels/Vikhr-7B-instruct_0.4 on the None dataset. It achieves the following results on the evaluation set:

• Loss: 2.5694

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: 0.0003
• train_batch_size: 1
• eval_batch_size: 1
• seed: 42
• gradient_accumulation_steps: 16
• 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: 10
• training_steps: 300

Training results

Training Loss	Epoch	Step	Validation Loss
2.8686	0.0010	1	2.9809
2.6271	0.0192	20	2.7117
2.4668	0.0384	40	2.7319
2.6359	0.0576	60	2.6860
2.7575	0.0768	80	2.6727
2.8165	0.0960	100	2.8388
2.5274	0.1152	120	2.6358
2.6776	0.1344	140	2.6357
2.5065	0.1536	160	2.6527
2.5421	0.1728	180	2.6056
2.6168	0.1920	200	2.6692
2.5725	0.2112	220	2.6160
2.6668	0.2304	240	2.5838
2.5096	0.2496	260	2.5708
2.4965	0.2688	280	2.5697
2.4599	0.2880	300	2.5694

Framework versions

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