File size: 2,610 Bytes
ea08385 |
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 |
---
datasets:
- iamtarun/python_code_instructions_18k_alpaca
language:
- en
metrics:
- code_eval
library_name: transformers
pipeline_tag: text-generation
tags:
- code
widget:
- text: 'def isprime(num):'
example_title: Code Example 1
- text: 'def factorial(num):'
example_title: Code Example 2
- text: 'def square(num):'
example_title: Code Example 3
---
# Competitive Programming LLM for Python Language
This model is a finetuned version of [codegen350M-mono](https://huggingface.co/Salesforce/codegen-350M-mono) on python code [dataset](https://huggingface.co/datasets/iamtarun/python_code_instructions_18k_alpaca) that uses alpaca style prompts while training.
## Prompt function
```python
'''
This function generates prompts using the problem description and input.
@param1 instruction: str - text problem description
@param2 inputs: str - input to the program
'''
def generate_prompt(instruction, inputs=""):
text = ("Below is an instruction that describes a task. Write a response that appropriately completes the request.\n\n"
"### Instruction:\n"
f"{instruction}\n\n"
"### Input:\n"
f"{inputs}\n\n"
"### Output:\n")
return text
```
## Usage
```python
from transformers import AutoModelForCausalLM, AutoTokenizer
# load model and tokenizer
model = AutoModelForCausalLM.from_pretrained("iamtarun/codegen-350M-mono-4bit-qlora", device_map="auto")
tokenizer = AutoTokenizer.from_pretrained("iamtarun/codegen-350M-mono-4bit-qlora")
# loading model for inference
model.eval()
# inference function
'''
This function takes text prompt as input which is generated from the generate_prompt function and returns the generated response
@param1 prompt: str - text prompt generated using generate_prompt function.
'''
def pipe(prompt):
device = "cuda"
inputs = tokenizer(prompt, return_tensors="pt").to(device)
with torch.no_grad():
output = model.generate(**inputs,
max_length=512,
do_sample=True,
temperature=0.5,
top_p=0.95,
repetition_penalty=1.15)
return tokenizer.decode(output[0].tolist(),
skip_special_tokens=True,
clean_up_tokenization_space=False)
# generating code for a problem description
instruction = "Write a function to calculate square of a number in python"
inputs = "number = 5"
prompt = generate_prompt(instruction, inputs)
print(pipe(prompt))
print("\n", "="*100)
``` |