See axolotl config
axolotl version: 0.4.0
# base_model: Qwen/Qwen-7B
base_model: Qwen/Qwen1.5-7B
model_type: AutoModelForCausalLM
tokenizer_type: AutoTokenizer
trust_remote_code: true
load_in_8bit: true
load_in_4bit: false
strict: false
datasets:
# - path: mhenrichsen/alpaca_2k_test
- path: /home/yujia/home/CN_Hateful/train_toxiCN_cn.json
# - path: /home/yujia/home/CN_Hateful/train_toxiCN.json
# - path: /home/yujia/home/CN_Hateful/train.json
# - path: /home/yujia/home/CN_Hateful/train_cn.json
ds_type: json
type: alpaca
dataset_prepared_path:
val_set_size: 0.05
output_dir: /home/yujia/home/CN_Hateful/trained_models/qwen/CN/toxi/1e-3/
# output_dir: /home/yujia/home/CN_Hateful/trained_models/qwen/toxi/1e-5/
# output_dir: /home/yujia/home/CN_Hateful/trained_models/qwen/cold/3e-4/
# output_dir: /home/yujia/home/CN_Hateful/trained_models/qwen/CN/cold/3e-4/
sequence_len: 256 # supports up to 8192
sample_packing: false
pad_to_sequence_len:
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:
wandb_entity:
wandb_watch:
wandb_name:
wandb_log_model:
gradient_accumulation_steps: 4
micro_batch_size: 2
num_epochs: 3
optimizer: adamw_bnb_8bit
lr_scheduler: cosine
learning_rate: 0.001
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:
warmup_steps: 10
evals_per_epoch: 4
eval_table_size:
eval_max_new_tokens: 20
saves_per_epoch: 1
debug:
deepspeed:
weight_decay: 0.0
fsdp:
fsdp_config:
special_tokens:
home/yujia/home/CN_Hateful/trained_models/qwen/CN/toxi/1e-3/
This model is a fine-tuned version of Qwen/Qwen1.5-7B on the None dataset. It achieves the following results on the evaluation set:
- Loss: 0.1333
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: 0.001
- train_batch_size: 2
- eval_batch_size: 2
- seed: 42
- distributed_type: multi-GPU
- num_devices: 2
- gradient_accumulation_steps: 4
- total_train_batch_size: 16
- total_eval_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: 3
Training results
Training Loss | Epoch | Step | Validation Loss |
---|---|---|---|
3.3182 | 0.0 | 1 | 3.3363 |
0.1432 | 0.25 | 142 | 1.5367 |
0.1667 | 0.5 | 284 | 0.1418 |
0.1329 | 0.75 | 426 | 0.1375 |
0.1372 | 1.0 | 568 | 0.1397 |
0.1369 | 1.26 | 710 | 0.1382 |
0.1537 | 1.51 | 852 | 0.1408 |
0.1204 | 1.76 | 994 | 0.1354 |
0.1262 | 2.01 | 1136 | 0.1343 |
0.12 | 2.26 | 1278 | 0.1380 |
0.146 | 2.51 | 1420 | 0.1335 |
0.1502 | 2.76 | 1562 | 0.1333 |
Framework versions
- PEFT 0.10.0
- Transformers 4.40.0.dev0
- Pytorch 2.2.1+cu121
- Datasets 2.18.0
- Tokenizers 0.15.0
- Downloads last month
- 2
Model tree for yujia23/axolotl-qwen-cn-1e-3-lora
Base model
Qwen/Qwen1.5-7B
Adapter
this model