103 lines
3.6 KiB
Markdown
103 lines
3.6 KiB
Markdown
---
|
|
library_name: transformers
|
|
tags:
|
|
- text-generation-inference
|
|
- Code
|
|
- Math
|
|
- RL
|
|
license: apache-2.0
|
|
language:
|
|
- en
|
|
base_model:
|
|
- Qwen/Qwen2.5-1.5B-Instruct
|
|
pipeline_tag: text-generation
|
|
---
|
|
|
|
|
|
|
|
# **Pictor-1338-QwenP-1.5B**
|
|
|
|
> **Pictor-1338-QwenP-1.5B** is a **code reasoning LLM** fine-tuned from **Qwen-1.5B** using **distributed reinforcement learning (RL)**. This model is designed to enhance coding proficiency, debugging accuracy, and step-by-step reasoning in software development tasks across multiple programming languages.
|
|
|
|
## **Key Features**
|
|
|
|
1. **Code Reasoning & Explanation**
|
|
Trained to **analyze, generate, and explain** code with a focus on logic, structure, and clarity. Supports functional, object-oriented, and procedural paradigms.
|
|
|
|
2. **Reinforcement Learning Fine-Tuning**
|
|
Enhanced using **distributed RL**, improving reward-aligned behavior in tasks like fixing bugs, completing functions, and understanding abstract instructions.
|
|
|
|
3. **Multi-Language Support**
|
|
Works fluently with **Python**, **JavaScript**, **C++**, and **Shell**, among others—ideal for general-purpose programming, scripting, and algorithmic tasks.
|
|
|
|
4. **Compact and Efficient**
|
|
At just **1.5B parameters**, it's lightweight enough for **edge deployments** and **developer tools** with strong reasoning capability.
|
|
|
|
5. **Debugging and Auto-Fix Capabilities**
|
|
Built to **identify bugs**, **recommend corrections**, and provide context-aware **explanations of issues** in codebases.
|
|
|
|
## **Quickstart with Transformers**
|
|
|
|
```python
|
|
from transformers import AutoModelForCausalLM, AutoTokenizer
|
|
|
|
model_name = "prithivMLmods/Pictor-1338-QwenP-1.5B"
|
|
|
|
model = AutoModelForCausalLM.from_pretrained(
|
|
model_name,
|
|
torch_dtype="auto",
|
|
device_map="auto"
|
|
)
|
|
tokenizer = AutoTokenizer.from_pretrained(model_name)
|
|
|
|
prompt = "Write a Python function that checks if a number is prime, and explain how it works."
|
|
|
|
messages = [
|
|
{"role": "system", "content": "You are a code reasoning assistant. Your job is to write correct code and explain the logic step-by-step."},
|
|
{"role": "user", "content": prompt}
|
|
]
|
|
text = tokenizer.apply_chat_template(
|
|
messages,
|
|
tokenize=False,
|
|
add_generation_prompt=True
|
|
)
|
|
model_inputs = tokenizer([text], return_tensors="pt").to(model.device)
|
|
|
|
generated_ids = model.generate(
|
|
**model_inputs,
|
|
max_new_tokens=512
|
|
)
|
|
generated_ids = [
|
|
output_ids[len(input_ids):] for input_ids, output_ids in zip(model_inputs.input_ids, generated_ids)
|
|
]
|
|
|
|
response = tokenizer.batch_decode(generated_ids, skip_special_tokens=True)[0]
|
|
```
|
|
|
|
## **Intended Use**
|
|
|
|
- **Code Assistance & IDE Integration**:
|
|
Smart autocomplete, bug detection, and function suggestion for developers.
|
|
|
|
- **Learning & Explanation**:
|
|
Ideal for **students** and **educators** in programming courses or interactive coding tutorials.
|
|
|
|
- **Automated Code Review & QA**:
|
|
Analyzes logic, structure, and potential bugs in code for quality assurance.
|
|
|
|
- **Edge & DevTool Deployments**:
|
|
Lightweight enough for browser extensions, local developer tools, and CLI-based assistants.
|
|
|
|
## **Limitations**
|
|
|
|
1. **Scaling Challenges**
|
|
May not handle large, complex codebases as well as larger models.
|
|
|
|
2. **Inconsistent Creativity**
|
|
May vary in performance for creative or unconventional coding tasks.
|
|
|
|
3. **Security Considerations**
|
|
Outputs should be audited to avoid insecure or vulnerable code patterns.
|
|
|
|
4. **Prompt Design Sensitivity**
|
|
Better output with clear instructions, function definitions, or examples. |