Hugging Face's logo

tpoisonooo
/
HuixiangDou-CR-LoRA-Qwen-32B

PEFT
Safetensors
qwen2
Generated from Trainer
Model card Files Community
HuixiangDou-CR-LoRA-Qwen-32B / README.md
tpoisonooo
init(project): update
fc2cafe
preview code
|
raw
history blame contribute delete
2.78 kB
metadata
library_name: peft
tags:
  - generated_from_trainer
base_model: Qwen/Qwen1.5-32B-Chat
model-index:
  - name: out-qwen32
    results: []

Built with Axolotl

See axolotl config

axolotl version: 0.4.0

base_model: /workspace/models/Qwen1.5-32B-Chat
trust_remote_code: true

load_in_8bit: false
load_in_4bit: false
strict: false

datasets:
  - path: /workspace/axolotl/alpaca.json
    type: alpaca
dataset_prepared_path:
val_set_size: 0.05
output_dir: ./out-qwen32


sequence_len: 1400  # supports up to 32k
sample_packing: false
pad_to_sequence_len: false

adapter: lora
lora_model_dir:
lora_r: 64
lora_alpha: 16
lora_dropout: 0.05
lora_target_linear: true
lora_fan_in_fan_out:

wandb_mode: online
wandb_project: huixiangdou-cr
wandb_entity:
wandb_watch:
wandb_name: qwen32
wandb_log_model:

gradient_accumulation_steps: 1
micro_batch_size: 4
num_epochs: 1
optimizer: paged_adamw_8bit
lr_scheduler: cosine
learning_rate: 0.0002

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

warmup_steps: 10
evals_per_epoch: 1
saves_per_epoch: 1
debug:
deepspeed:
weight_decay: 0.0
fsdp:
fsdp_config:
special_tokens:

out-qwen32

This model was trained from scratch on the None dataset. It achieves the following results on the evaluation set:

  • Loss: 0.0370

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: 4
  • eval_batch_size: 4
  • seed: 42
  • distributed_type: multi-GPU
  • num_devices: 2
  • total_train_batch_size: 8
  • total_eval_batch_size: 8
  • optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
  • lr_scheduler_type: cosine
  • lr_scheduler_warmup_steps: 10
  • num_epochs: 1

Training results

Training Loss Epoch Step Validation Loss
0.0183 1.0 273 0.0370

Framework versions

  • PEFT 0.10.0
  • Transformers 4.40.0.dev0
  • Pytorch 2.1.2+cu121
  • Datasets 2.15.0
  • Tokenizers 0.15.0