Hugging Face's logo

ResplendentAI
/
Qwen_jeiku_LoRA_128

PEFT
qwen2
Generated from Trainer
4-bit precision
bitsandbytes
Model card Files Community
Qwen_jeiku_LoRA_128 / README.md
jeiku's picture
jeiku
Upload 10 files
e0b9251 verified
preview code
|
raw
history blame contribute delete
3.24 kB
metadata
license: apache-2.0
library_name: peft
tags:
  - generated_from_trainer
base_model: Qwen/Qwen2-7B
model-index:
  - name: outputs/out
    results: []

Built with Axolotl

See axolotl config

axolotl version: 0.4.1

adapter: qlora
base_model: Qwen/Qwen2-7B
bf16: auto
dataset_prepared_path: null
datasets:
- path: ResplendentAI/Jeiku
  type: completion
debug: null
deepspeed: null
early_stopping_patience: null
eval_sample_packing: false
evals_per_epoch: 2
flash_attention: true
fp16: null
gradient_accumulation_steps: 4
gradient_checkpointing: true
gradient_checkpointing_kwargs:
  use_reentrant: false
group_by_length: false
learning_rate: 2.0e-05
load_in_4bit: true
load_in_8bit: false
local_rank: null
logging_steps: 1
lora_alpha: 128
lora_dropout: 0.05
lora_fan_in_fan_out: null
lora_model_dir: null
lora_r: 128
lora_target_linear: true
lr_scheduler: cosine
micro_batch_size: 2
num_epochs: 4
optimizer: adamw_torch
output_dir: ./outputs/out
pad_to_sequence_len: false
resume_from_checkpoint: null
sample_packing: false
saves_per_epoch: 1
sequence_len: 8192
special_tokens: null
strict: false
tf32: true
train_on_inputs: false
trust_remote_code: true
val_set_size: 0.05
wandb_entity: null
wandb_log_model: null
wandb_name: null
wandb_project: null
wandb_watch: null
warmup_steps: 10
weight_decay: 0.0
xformers_attention: null

outputs/out

This model is a fine-tuned version of Qwen/Qwen2-7B on the None dataset. It achieves the following results on the evaluation set:

  • Loss: 2.8811

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: 2e-05
  • train_batch_size: 2
  • eval_batch_size: 2
  • seed: 42
  • gradient_accumulation_steps: 4
  • total_train_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: 4

Training results

Training Loss Epoch Step Validation Loss
6.5037 0.0014 1 4.0893
3.3749 0.4998 366 3.0970
4.2647 0.9997 732 2.9672
2.2511 1.4995 1098 2.9058
2.5316 1.9993 1464 2.8220
2.2853 2.4991 1830 2.8457
2.2372 2.9990 2196 2.8423
2.5647 3.4988 2562 2.8807
2.7001 3.9986 2928 2.8811

Framework versions

  • PEFT 0.11.1
  • Transformers 4.41.1
  • Pytorch 2.1.2+cu121
  • Datasets 2.19.1
  • Tokenizers 0.19.1