File size: 2,697 Bytes
a93c45e 9e77dc2 a93c45e 9bc7370 a93c45e 9e77dc2 a93c45e 9bc7370 a93c45e 9e77dc2 a93c45e 9e77dc2 a93c45e 9e77dc2 a93c45e 9bc7370 a93c45e |
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 61 62 63 64 65 66 67 68 69 70 71 72 73 74 75 76 77 78 79 80 81 82 83 84 85 86 87 88 89 90 91 92 93 |
---
license: mit
base_model: TheBloke/zephyr-7B-beta-GPTQ
tags:
- generated_from_trainer
model-index:
- name: KUETLLM_zephyr
results: []
---
KUETLLM is a [zephyr7b-beta](https://huggingface.co/HuggingFaceH4/zephyr-7b-beta) finetune, using a dataset with prompts and answers about Khulna University of Engineering and Technology.
It was loaded in 8 bit quantization using [bitsandbytes](https://github.com/TimDettmers/bitsandbytes). [LORA](https://huggingface.co/docs/diffusers/main/en/training/lora) was used to finetune an adapter, which was leter merged with the base unquantized model.
<!-- This model card has been generated automatically according to the information the Trainer had access to. You
should probably proofread and complete it, then remove this comment. -->
# KUETLLM_zephyr
This model is a fine-tuned version of [TheBloke/zephyr-7B-beta-GPTQ](https://huggingface.co/TheBloke/zephyr-7B-beta-GPTQ) on the None dataset.
## Model description
Below is the training configuarations for the finetuning process:
```
LoraConfig:
r=16,
lora_alpha=16,
target_modules=["q_proj", "v_proj","k_proj","o_proj","gate_proj","up_proj","down_proj"],
lora_dropout=0.05,
bias="none",
task_type="CAUSAL_LM"
```
```
TrainingArguments:
per_device_train_batch_size=12,
gradient_accumulation_steps=1,
optim='paged_adamw_8bit',
learning_rate=5e-06 ,
fp16=True,
logging_steps=10,
num_train_epochs = 1,
output_dir=zephyr_lora_output,
remove_unused_columns=False,
```
## 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: 24
- eval_batch_size: 8
- seed: 42
- gradient_accumulation_steps: 4
- total_train_batch_size: 96
- optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
- lr_scheduler_type: linear
- num_epochs: 2
- mixed_precision_training: Native AMP
### Inference
```
def process_data_sample(example):
processed_example = "<|system|>\nYou are a KUET authority managed chatbot, help users by answering their queries about KUET.\n<|user|>\n" + example + "\n<|assistant|>\n"
return processed_example
inp_str = process_data_sample("Tell me about KUET.")
inputs = tokenizer(inp_str, return_tensors="pt")
generation_config = GenerationConfig(
do_sample=True,
top_k=1,
temperature=0.1,
max_new_tokens=256,
pad_token_id=tokenizer.eos_token_id
)
outputs = model.generate(**inputs, generation_config=generation_config)
print(tokenizer.decode(outputs[0], skip_special_tokens=True))
```
### Framework versions
- Transformers 4.36.0.dev0
- Pytorch 2.1.1+cu121
- Datasets 2.15.0
- Tokenizers 0.15.0
|