See axolotl config

axolotl version: 0.4.1

adapter: lora
auto_find_batch_size: true
base_model: bigscience/bloom-560m
bf16: auto
chat_template: llama3
dataloader_num_workers: 12
dataset_prepared_path: null
datasets:
- data_files:
  - 556019029e37cf14_train_data.json
  ds_type: json
  format: custom
  path: /workspace/input_data/556019029e37cf14_train_data.json
  type:
    field_input: ''
    field_instruction: instruction
    field_output: output
    format: '{instruction}'
    no_input_format: '{instruction}'
    system_format: '{system}'
    system_prompt: ''
debug: null
deepspeed: null
early_stopping_patience: 3
early_stopping_threshold: 0.001
eval_max_new_tokens: 128
eval_steps: 40
flash_attention: false
fp16: null
fsdp: null
fsdp_config: null
gradient_accumulation_steps: 2
gradient_checkpointing: false
group_by_length: false
hub_model_id: mrferr3t/51afafae-58d5-4667-8278-1b2974a28289
hub_repo: null
hub_strategy: checkpoint
hub_token: null
learning_rate: 0.0003
load_in_4bit: false
load_in_8bit: false
local_rank: null
logging_steps: 100
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
micro_batch_size: 32
mlflow_experiment_name: /tmp/556019029e37cf14_train_data.json
model_type: AutoModelForCausalLM
num_epochs: 50
optimizer: adamw_bnb_8bit
output_dir: miner_id_24
pad_to_sequence_len: true
s2_attention: null
sample_packing: false
save_steps: 40
saves_per_epoch: 0
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: e189dd85-ea36-46aa-8f01-f41e7f68678d
wandb_project: Gradients-On-Demand
wandb_run: your_name
wandb_runid: e189dd85-ea36-46aa-8f01-f41e7f68678d
warmup_ratio: 0.05
weight_decay: 0.0
xformers_attention: null

51afafae-58d5-4667-8278-1b2974a28289

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

• Loss: 3.1195

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: 32
• eval_batch_size: 32
• seed: 42
• gradient_accumulation_steps: 2
• total_train_batch_size: 64
• optimizer: Use adamw_bnb_8bit 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: 179
• num_epochs: 50

Training results

Training Loss	Epoch	Step	Validation Loss
No log	0.0069	1	3.6892
No log	0.2778	40	3.4418
No log	0.5556	80	3.3057
6.847	0.8333	120	3.2433
6.847	1.1111	160	3.2219
6.3617	1.3889	200	3.1796
6.3617	1.6667	240	3.1717
6.3617	1.9444	280	3.1511
6.1539	2.2222	320	3.1385
6.1539	2.5	360	3.1462
5.8916	2.7778	400	3.1116
5.8916	3.0556	440	3.1356
5.8916	3.3333	480	3.1250
5.6539	3.6111	520	3.1195

Framework versions

• PEFT 0.13.2
• Transformers 4.46.0
• Pytorch 2.3.1+cu121
• Datasets 3.0.1
• Tokenizers 0.20.1