QuixiAI/Ina-v11.1-AWQ

See axolotl config

axolotl version: 0.4.1

base_model: NeverSleep/Lumimaid-v0.2-70B
model_type: LlamaForCausalLM
tokenizer_type: AutoTokenizer  # PreTrainedTokenizerFast

load_in_8bit: false
load_in_4bit: true
strict: false

chat_template: llama3
rl: dpo
datasets:
  - path: theIndividual/dpo_pairs_ina_markdown_expanded
    type: chat_template.default
    chat_template: llama3
    field_messages: conversation
    field_chosen: chosen
    field_rejected: rejected
    message_field_role: role
    message_field_content: content
    roles:
      system:
        - system
      user:
        - user
      assistant:
        - assistant

adapter: qlora
lora_model_dir:

output_dir: llama3-markdown-dpo

sequence_len: 3096
sample_packing: false
pad_to_sequence_len: true

lora_r: 16
lora_alpha: 16
lora_dropout: 0.05
lora_target_modules:
lora_target_linear: true
lora_fan_in_fan_out:

wandb_project: llama3-markdown-dpo
wandb_entity:
wandb_watch:
wandb_name:
wandb_log_model:

gradient_accumulation_steps: 4
micro_batch_size: 2
num_epochs: 4
optimizer: adamw_torch
lr_scheduler: cosine
learning_rate: 0.00003

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

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

warmup_steps: 5
evals_per_epoch: 4
eval_table_size:
saves_per_epoch: 1
debug:
deepspeed:
weight_decay: 0.0
fsdp:
  - full_shard
  - auto_wrap
fsdp_config:
  fsdp_auto_wrap_policy: TRANSFORMER_BASED_WRAP
  fsdp_backward_prefetch: BACKWARD_PRE
  fsdp_cpu_ram_efficient_loading: true
  fsdp_forward_prefetch: true
  fsdp_offload_params: true
  fsdp_sharding_strategy: FULL_SHARD
  fsdp_state_dict_type: FULL_STATE_DICT
  fsdp_sync_module_states: true
  fsdp_use_orig_params: false
special_tokens:
  pad_token: <|end_of_text|>

llama3-markdown-dpo

This model is a fine-tuned version of NeverSleep/Lumimaid-v0.2-70B on an unknown 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: 3e-05
• train_batch_size: 2
• eval_batch_size: 8
• seed: 42
• distributed_type: multi-GPU
• num_devices: 2
• gradient_accumulation_steps: 4
• total_train_batch_size: 16
• total_eval_batch_size: 16
• optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
• lr_scheduler_type: cosine
• lr_scheduler_warmup_steps: 5
• training_steps: 413

Training results

Framework versions

• PEFT 0.12.0
• Transformers 4.44.0
• Pytorch 2.3.1+cu121
• Datasets 2.20.0
• Tokenizers 0.19.1