Hugging Face's logo

kaist-ai
/
janus-orpo-7b

Text Generation
Transformers
PyTorch
Safetensors
English
mistral
axolotl
Generated from Trainer
orpo
conversational
text-generation-inference
Inference Endpoints
Model card Files Community
janus-orpo-7b / README.md
Seongyun's picture
Seongyun
End of training
aeab9f0 verified
preview code
|
raw
history blame
2.96 kB
metadata
license: apache-2.0
base_model: alpindale/Mistral-7B-v0.2-hf
tags:
  - axolotl
  - generated_from_trainer
model-index:
  - name: mpa-Mistral-7b-v0.2-hf-sft-66k-orpo
    results: []

Built with Axolotl

See axolotl config

axolotl version: 0.4.0

base_model: alpindale/Mistral-7B-v0.2-hf
model_type: MistralForCausalLM
tokenizer_type: LlamaTokenizer

load_in_8bit: false
load_in_4bit: false
strict: false

rl: orpo
orpo_alpha: 0.1
remove_unused_columns: false
chat_template: chatml
datasets:
  - path: /mnt/nas/seongyun/MPA/data/training/preferences_v1_responses_for_orpo_64k_v2.jsonl
    type: orpo.chat_template
    # conversation: mistral
      
dataset_prepared_path:
hub_model_id: kaist-ai/mpa-Mistral-7b-v0.2-hf-sft-66k-orpo
hub_strategy: checkpoint
# val_set_size: 0
output_dir: /mnt/nas/seongyun/axolotl/outputs/mpa_66k_orpo

sequence_len: 4096
sample_packing: false
pad_to_sequence_len: true
eval_sample_packing: false

wandb_project: mpa
wandb_entity: seongyun
wandb_watch:
wandb_name: mpa_mistral-7b-v0.2-hf-66k-orpo
wandb_log_model:

gradient_accumulation_steps: 4
micro_batch_size: 1
num_epochs: 2
optimizer: adamw_bnb_8bit
lr_scheduler: cosine
learning_rate: 0.000005

train_on_inputs: false
group_by_length: false
bf16: auto
fp16:
tf32: false

gradient_checkpointing: true
early_stopping_patience:
resume_from_checkpoint:
local_rank:
logging_steps: 1
xformers_attention:
flash_attention: true

warmup_steps: 10
# evals_per_epoch: 4
eval_table_size:
# eval_max_new_tokens: 128
saves_per_epoch: 1
debug:
deepspeed:
weight_decay: 0.0
fsdp:
fsdp_config:
special_tokens:

mpa-Mistral-7b-v0.2-hf-sft-66k-orpo

This model is a fine-tuned version of alpindale/Mistral-7B-v0.2-hf on the None dataset.

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-06
  • train_batch_size: 1
  • eval_batch_size: 1
  • seed: 42
  • distributed_type: multi-GPU
  • num_devices: 4
  • gradient_accumulation_steps: 4
  • total_train_batch_size: 16
  • total_eval_batch_size: 4
  • optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
  • lr_scheduler_type: cosine
  • lr_scheduler_warmup_steps: 10
  • num_epochs: 2

Training results

Framework versions

  • Transformers 4.40.0.dev0
  • Pytorch 2.2.2
  • Datasets 2.18.0
  • Tokenizers 0.15.0