amphora/l3kpm-dpo-lora

See axolotl config

axolotl version: 0.4.1

base_model: amphora/l3kpm-96
model_type: LlamaForCausalLM
tokenizer_type: AutoTokenizer

load_in_8bit: false
load_in_4bit: false
strict: false

rl: dpo
chat_template: llama3
datasets:
  - path: amphora/kpmDPO
    split: train
    type: chatml.argilla

# datasets:
#   - path: a
#     type: chat_template
#     chat_template: llama3
#     field_messages: messages
#     message_field_role: role
#     message_field_content: content
#     roles:
#       user:
#         - user
#       assistant:
#         - assistant
val_set_size: 0.01
output_dir: ./outputs/lora-out

sequence_len: 4096
sample_packing: false
pad_to_sequence_len: false

adapter: lora
lora_model_dir:
lora_r: 32
lora_alpha: 16
lora_dropout: 0.05
lora_target_linear: true
lora_fan_in_fan_out:

wandb_project: fincode
wandb_entity:
wandb_watch:
wandb_name:
wandb_log_model:

gradient_accumulation_steps: 32
micro_batch_size: 1
num_epochs: 1
optimizer: adamw_bnb_8bit
lr_scheduler: cosine
learning_rate: 0.0002

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
s2_attention:
deepspeed: deepspeed_configs/zero3.json

warmup_steps: 10
evals_per_epoch: 4
eval_table_size:
eval_max_new_tokens: 128
saves_per_epoch: 2
debug:
deepspeed:
weight_decay: 0.0
fsdp:
fsdp_config:
special_tokens:
  pad_token: <|end_of_text|>

outputs/lora-out

This model is a fine-tuned version of amphora/l3kpm-96 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: 0.0002
• train_batch_size: 1
• eval_batch_size: 8
• seed: 42
• distributed_type: multi-GPU
• num_devices: 8
• gradient_accumulation_steps: 32
• total_train_batch_size: 256
• total_eval_batch_size: 64
• optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
• lr_scheduler_type: cosine
• lr_scheduler_warmup_steps: 10
• training_steps: 20

Training results

Framework versions

• PEFT 0.11.1
• Transformers 4.42.3
• Pytorch 2.1.2
• Datasets 2.19.1
• Tokenizers 0.19.1