jayzeng/llm-fun

See axolotl config

axolotl version: 0.4.0

base_model: Qwen/Qwen1.5-MoE-A2.7B
trust_remote_code: true

load_in_8bit: false
load_in_4bit: true
strict: false

datasets:
  - path: mhenrichsen/alpaca_2k_test
    type: alpaca
dataset_prepared_path:
val_set_size: 0.05
output_dir: ./out

sequence_len: 1024  # supports up to 32k
sample_packing: false
pad_to_sequence_len: false

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

wandb_project:
wandb_entity:
wandb_watch:
wandb_name:
wandb_log_model:

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

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

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: 4
saves_per_epoch: 1
debug:
deepspeed:
weight_decay: 0.0
fsdp:
fsdp_config:
special_tokens:

out

This model is a fine-tuned version of Qwen/Qwen1.5-MoE-A2.7B on the None dataset. It achieves the following results on the evaluation set:

• Loss: 1.2553

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: 1
• seed: 42
• gradient_accumulation_steps: 4
• total_train_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: 4

Training results

Training Loss	Epoch	Step	Validation Loss
0.8629	0.0	1	0.9370
0.6917	0.25	119	0.8805
0.9783	0.5	238	0.8783
0.9578	0.75	357	0.8827
0.4772	1.0	476	0.8900
0.4653	1.25	595	0.9620
0.5907	1.5	714	0.9532
0.7364	1.75	833	0.9360
0.2611	2.0	952	0.9570
0.1999	2.25	1071	1.0415
0.1532	2.51	1190	1.0776
0.0455	2.76	1309	1.0920
0.087	3.01	1428	1.1094
0.0183	3.26	1547	1.2266
0.0135	3.51	1666	1.2604
0.1929	3.76	1785	1.2553

Framework versions

• PEFT 0.10.0
• Transformers 4.40.0.dev0
• Pytorch 2.1.2+cu118
• Datasets 2.18.0
• Tokenizers 0.15.0