Model Card for amit-s-agrahari-coder-lora-v1

This is a LoRA fine-tuned version of the deepseek-ai/deepseek-coder-1.3b-base model, created to generate C programming solutions for algorithmic problems.
It was trained using PEFT (Parameter-Efficient Fine-Tuning) on a curated set of C programming tasks.

---

Model Details

Model Description

This is a parameter-efficient fine-tuned model based on DeepSeek Coder.
It focuses on generating high-quality, compilable C code for algorithmic and structured programming problems.

• Developed by: BlackIIIWhite
• Funded by [optional]: N/A
• Shared by [optional]: BlackIIIWhite
• Model type: Causal Language Model (LoRA fine-tuned)
• Language(s): Primarily C code generation
• License: MIT
• Finetuned from model: deepseek-ai/deepseek-coder-1.3b-base

Model Sources

• Repository: BlackIIIWhite/amit-s-agrahari-coder-lora-v1
• Paper: N/A
• Demo: Coming soon

---

Uses

Direct Use

This model can be directly used for:

• Generating C programming solutions for algorithmic challenges
• Code completion and function generation
• Educational purposes to demonstrate LoRA fine-tuning

Downstream Use

• Further fine-tuning for other programming languages
• Integration in code-assistant applications

Out-of-Scope Use

• Deploying in production environments without human code review
• Security-critical or safety-critical applications
• Generating sensitive or proprietary code without verification

---

Bias, Risks, and Limitations

The model may:

• Produce incorrect or unoptimized code
• Miss edge cases
• Reflect biases present in its training data

Recommendations

Always review and test generated code. This model is for educational and research purposes.

---

How to Get Started with the Model

from transformers import AutoTokenizer, AutoModelForCausalLM
from peft import PeftModel

base_model = "deepseek-ai/deepseek-coder-1.3b-base"
adapter_model = "BlackIIIWhite/amit-s-agrahari-coder-lora-v1"

tokenizer = AutoTokenizer.from_pretrained(base_model)
model = AutoModelForCausalLM.from_pretrained(base_model)
model = PeftModel.from_pretrained(model, adapter_model)

prompt = "Write a C program to calculate factorial of a number."
inputs = tokenizer(prompt, return_tensors="pt")
outputs = model.generate(**inputs, max_new_tokens=150)
print(tokenizer.decode(outputs[0], skip_special_tokens=True))