metadata
base_model: Qwen/CodeQwen1.5-7B-Chat
library_name: peft
license: other
tags:
- generated_from_trainer
model-index:
- name: workspace/axolotl/vinh/Qwen_CodeQwen1.5-7B-Chat-lora-2024-07-01-02-04-03
results: []
See axolotl config
axolotl version: 0.4.1
base_model: Qwen/CodeQwen1.5-7B-Chat
model_type: AutoModelForCausalLM
tokenizer_type: AutoTokenizer
load_in_8bit: false
load_in_4bit: false
strict: false
datasets:
- path: /workspace/axolotl/vinh/PAL/input_output_qwen.json
type: input_output
- path: /workspace/axolotl/vinh/INSTRUCT/input_output_qwen.json
type: input_output
dataset_prepared_path:
val_set_size: 0.05
eval_sample_packing: false
output_dir: /workspace/axolotl/vinh/Qwen_CodeQwen1.5-7B-Chat-lora-2024-07-01-02-04-03
sequence_len: 2048
sample_packing: false
pad_to_sequence_len: false
adapter: lora
lora_model_dir:
lora_r: 64
lora_alpha: 128
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: 128
micro_batch_size: 1
num_epochs: 3
optimizer: paged_adamw_32bit
lr_scheduler: cosine
learning_rate: 2e-4
train_on_inputs: false
group_by_length: false
bf16: auto
fp16:
tf32: false
gradient_checkpointing: false
early_stopping_patience:
resume_from_checkpoint:
local_rank:
logging_steps: 1
xformers_attention:
flash_attention: true
s2_attention:
loss_watchdog_threshold: 5.0
loss_watchdog_patience: 3
warmup_steps: 10
evals_per_epoch: 10
eval_table_size:
eval_max_new_tokens: 512
saves_per_epoch: 2
save_total_limit: 20
debug:
deepspeed:
weight_decay: 0.0
fsdp:
fsdp_config:
workspace/axolotl/vinh/Qwen_CodeQwen1.5-7B-Chat-lora-2024-07-01-02-04-03
This model is a fine-tuned version of Qwen/CodeQwen1.5-7B-Chat on the None dataset. It achieves the following results on the evaluation set:
- Loss: 0.1189
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: 128
- total_train_batch_size: 128
- optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
- lr_scheduler_type: cosine
- lr_scheduler_warmup_steps: 10
- num_epochs: 3
Training results
| Training Loss | Epoch | Step | Validation Loss |
|---|---|---|---|
| 0.6497 | 0.0063 | 1 | 0.5999 |
| 0.2252 | 0.1011 | 16 | 0.2502 |
| 0.2024 | 0.2023 | 32 | 0.2020 |
| 0.1581 | 0.3034 | 48 | 0.1804 |
| 0.1912 | 0.4045 | 64 | 0.1682 |
| 0.1692 | 0.5056 | 80 | 0.1580 |
| 0.1401 | 0.6068 | 96 | 0.1516 |
| 0.1204 | 0.7079 | 112 | 0.1463 |
| 0.1336 | 0.8090 | 128 | 0.1420 |
| 0.1339 | 0.9101 | 144 | 0.1380 |
| 0.101 | 1.0113 | 160 | 0.1346 |
| 0.0871 | 1.1124 | 176 | 0.1330 |
| 0.1035 | 1.2135 | 192 | 0.1320 |
| 0.1025 | 1.3146 | 208 | 0.1300 |
| 0.0936 | 1.4158 | 224 | 0.1263 |
| 0.0797 | 1.5169 | 240 | 0.1241 |
| 0.1014 | 1.6180 | 256 | 0.1220 |
| 0.0984 | 1.7191 | 272 | 0.1196 |
| 0.1078 | 1.8203 | 288 | 0.1184 |
| 0.0803 | 1.9214 | 304 | 0.1171 |
| 0.0658 | 2.0225 | 320 | 0.1164 |
| 0.0517 | 2.1236 | 336 | 0.1214 |
| 0.0598 | 2.2248 | 352 | 0.1203 |
| 0.0704 | 2.3259 | 368 | 0.1198 |
| 0.0787 | 2.4270 | 384 | 0.1192 |
| 0.0537 | 2.5281 | 400 | 0.1190 |
| 0.0547 | 2.6293 | 416 | 0.1189 |
| 0.0589 | 2.7304 | 432 | 0.1189 |
| 0.0655 | 2.8315 | 448 | 0.1190 |
| 0.0613 | 2.9326 | 464 | 0.1189 |
Framework versions
- PEFT 0.11.1
- Transformers 4.41.1
- Pytorch 2.1.2+cu118
- Datasets 2.19.1
- Tokenizers 0.19.1