metadata

base_model: unsloth/Qwen2.5-Coder-7B-Instruct-bnb-4bit
library_name: peft
pipeline_tag: text-generation
tags:
  - base_model:adapter:unsloth/Qwen2.5-Coder-7B-Instruct-bnb-4bit
  - lora
  - sft
  - transformers
  - trl
  - unsloth
license: apache-2.0
datasets:
  - open-r1/codeforces-cots

Model Card for Model ID

Model Card for SaffalPoosh/reasoning_cpp_llm

This is a QLoRA adapter trained on C++ coding tasks and designed for reasoning-based code generation. The model specializes in solving algorithmic problems with step-by-step reasoning and generating optimized C++ solutions.

Example Usage

Problem Example

example_problem = """
A robot is situated at the top-left corner of an m x n grid. The robot can only move either down or right at any point in time. It wants to reach the bottom-right corner of the grid. Some cells in the grid are blocked by obstacles. How many unique paths can the robot take to reach the destination?

Constraints:
Time limit per test: 2.0 seconds
Memory limit per test: 256.0 megabytes
1 ≤ m, n ≤ 100
Grid cells are either 0 (empty) or 1 (obstacle).

Input Format:
The first line contains two integers m and n — the dimensions of the grid.
The next m lines each contain n integers (0 or 1) representing the grid.

Output Format:
Print a single integer — the number of unique paths.

Example:
Input:
3 3
0 0 0
0 1 0
0 0 0
"""

Model Loading and Inference

from unsloth import FastLanguageModel
from transformers import TextStreamer
from transformers import TextIteratorStreamer
from threading import Thread

# Model configuration
model_path = "SaffalPoosh/reasoning_cpp_llm"
max_seq_length = 16000 
dtype = None 
load_in_4bit = True 

# Load model and tokenizer
model, tokenizer = FastLanguageModel.from_pretrained(
    model_name=model_path,
    max_seq_length=max_seq_length, 
    dtype=dtype,        
    load_in_4bit=load_in_4bit,
    local_files_only=False
)

# This will download the base model and then patch by applying the LoRA adapters
FastLanguageModel.for_inference(model)

# Prepare Input Data
input_text = example_problem
inputs = tokenizer(input_text, return_tensors="pt")
inputs = {k: v.to("cuda") for k, v in inputs.items()}

# Initialize the text streamer
text_streamer = TextIteratorStreamer(tokenizer, skip_special_tokens=False)

# Perform Inference with streaming
stream_catcher = Thread(
    target=model.generate, 
    kwargs={
        **inputs, 
        "do_sample": True, 
        "streamer": text_streamer, 
        "max_new_tokens": 10000
    }
)

stream_catcher.start()

# Stream output to console and file
with open("output.txt", "w") as f:
    for token in text_streamer:
        print(token, end="", flush=True)
        f.write(token)

stream_catcher.join()

Model Details

Model Type: QLoRA Fine-tuned Language Model
Base Model: [Specify base model if known]
Training Focus: C++ algorithmic problem solving with reasoning
Max Sequence Length: 16,000 tokens
Quantization: 4-bit loading supported
Hardware Requirements: CUDA-compatible GPU recommended

Training Details

Training Method: QLoRA (Quantized Low-Rank Adaptation)
Dataset: C++ coding tasks with reasoning annotations
Task Type: Code generation with step-by-step reasoning
Optimization: Focused on algorithmic problem solving

Usage Notes

The model generates reasoning-based solutions for C++ programming problems
Supports streaming inference for real-time output
The output.txt file contains the complete generated solution
Designed to handle competitive programming style problems with constraints

Output Format

The model typically generates:

Problem analysis and reasoning
Algorithm explanation
Complete C++ implementation
Time and space complexity analysis

Requirements

pip install unsloth transformers torch

Hardware Requirements

GPU: CUDA-compatible GPU (recommended)
Memory: Sufficient VRAM for 4-bit quantized model
Storage: Space for base model download and adapter weights

Model Details

Model Description

Developed by: [More Information Needed]
Funded by [optional]: [More Information Needed]
Shared by [optional]: [More Information Needed]
Model type: [More Information Needed]
Language(s) (NLP): [More Information Needed]
License: [More Information Needed]
Finetuned from model [optional]: [More Information Needed]

Model Sources [optional]

Repository: [More Information Needed]
Paper [optional]: [More Information Needed]
Demo [optional]: [More Information Needed]

Uses

Direct Use

[More Information Needed]

Downstream Use [optional]

[More Information Needed]

Out-of-Scope Use

[More Information Needed]

Bias, Risks, and Limitations

[More Information Needed]

Recommendations

Users (both direct and downstream) should be made aware of the risks, biases and limitations of the model. More information needed for further recommendations.

How to Get Started with the Model

Use the code below to get started with the model.

[More Information Needed]

Training Details

Training Data

[More Information Needed]

Training Procedure

Preprocessing [optional]

[More Information Needed]

Training Hyperparameters

Training regime: [More Information Needed]

Speeds, Sizes, Times [optional]

[More Information Needed]

Evaluation

Testing Data, Factors & Metrics

Testing Data

[More Information Needed]

Factors

[More Information Needed]

Metrics

[More Information Needed]

Results

[More Information Needed]

Summary

Model Examination [optional]

[More Information Needed]

Environmental Impact

Carbon emissions can be estimated using the Machine Learning Impact calculator presented in Lacoste et al. (2019).

Hardware Type: [More Information Needed]
Hours used: [More Information Needed]
Cloud Provider: [More Information Needed]
Compute Region: [More Information Needed]
Carbon Emitted: [More Information Needed]

Technical Specifications [optional]

Model Architecture and Objective

[More Information Needed]

Compute Infrastructure

[More Information Needed]

Hardware

[More Information Needed]

Software

[More Information Needed]

Citation [optional]

BibTeX:

[More Information Needed]

APA:

[More Information Needed]

Glossary [optional]

[More Information Needed]

More Information [optional]

[More Information Needed]

Model Card Authors [optional]

[More Information Needed]

Model Card Contact

[More Information Needed]

Framework versions

PEFT 0.17.1