See axolotl config

axolotl version: 0.4.1

adapter: lora
base_model: bigscience/bloomz-560m
bf16: true
chat_template: llama3
datasets:
- data_files:
  - 77f692c8c486c799_train_data.json
  ds_type: json
  format: custom
  path: /workspace/input_data/77f692c8c486c799_train_data.json
  type:
    field_instruction: ru_text
    field_output: text
    format: '{instruction}'
    no_input_format: '{instruction}'
    system_format: '{system}'
    system_prompt: ''
debug: null
deepspeed: null
early_stopping_patience: 2
eval_batch_size: 2
eval_max_new_tokens: 128
eval_steps: 5
flash_attention: false
fp16: false
fsdp: null
fsdp_config: null
gradient_accumulation_steps: 4
gradient_checkpointing: true
group_by_length: false
hub_model_id: sn56a1/ab1a0f30-07df-4fb0-8fac-78b2ccb5b20e
hub_repo: null
hub_strategy: checkpoint
hub_token: null
learning_rate: 5.0e-05
load_in_4bit: false
load_in_8bit: false
local_rank: null
logging_steps: 1
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: 200
micro_batch_size: 2
mlflow_experiment_name: /tmp/77f692c8c486c799_train_data.json
model_type: AutoModelForCausalLM
num_epochs: 1
optimizer: adamw_bnb_8bit
optimizer_betas:
- 0.9
- 0.999
optimizer_epsilon: 1e-08
output_dir: miner_id_24
pad_to_sequence_len: true
resume_from_checkpoint: null
s2_attention: null
sample_packing: false
save_steps: 10
seed: 2145909994
sequence_len: 512
shuffle: true
strict: false
tf32: false
tokenizer_type: AutoTokenizer
torch_compile: true
train_on_inputs: false
trust_remote_code: true
val_set_size: 0.05
wandb_entity: sn56-miner
wandb_mode: disabled
wandb_name: null
wandb_project: god
wandb_run: 0fmo
wandb_runid: null
warmup_steps: 5
weight_decay: 0.0
xformers_attention: null

ab1a0f30-07df-4fb0-8fac-78b2ccb5b20e

This model is a fine-tuned version of bigscience/bloomz-560m on the None dataset. It achieves the following results on the evaluation set:

• Loss: 4.5631

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: 2
• eval_batch_size: 2
• seed: 2145909994
• distributed_type: multi-GPU
• num_devices: 4
• gradient_accumulation_steps: 4
• total_train_batch_size: 32
• total_eval_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: 5
• training_steps: 200

Training results

Training Loss	Epoch	Step	Validation Loss
17.5898	0.0002	1	4.5629
17.7344	0.0008	5	4.5628
17.2773	0.0016	10	4.5610
16.957	0.0024	15	4.5635
19.2695	0.0032	20	4.5631

Framework versions

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