README.md

---
language:
  - en
license: mit
base_model: moonshotai/Kimi-K2-Instruct
tags:
  - coding
  - reasoning
  - winter-llm
  - agents
  - math
creators:
  - name: INEZA AIME BRUNO
    url: https://huggingface.co/brucoder
---

<div align="center">

# ❄️ Winter Core 1.2 Code

**The strongest model in the Winter LLM family**

*Built by [INEZA AIME BRUNO](https://huggingface.co/brucoder)*

[![License: MIT](https://img.shields.io/badge/License-MIT-blue.svg)](https://opensource.org/licenses/MIT)
[![HuggingFace](https://img.shields.io/badge/🤗-brucoder-yellow)](https://huggingface.co/brucoder)
[![Winter LLM](https://img.shields.io/badge/Winter-LLM%20Family-lightblue)](https://huggingface.co/brucoder)

</div>

---

## What is Winter Core 1.2 Code?

**Winter Core 1.2 Code** is the flagship model of the Winter LLM project, created by **INEZA AIME BRUNO**. It is built for people who need a model that does not just answer — it *thinks*, it *codes*, and it *solves*.

It excels at:

- **Code generation** — writes clean, working code in Python, JavaScript, TypeScript, Rust, Go, C++, SQL, and more
- **Debugging** — finds bugs, explains why they happen, and fixes them
- **Reasoning** — breaks down complex problems step by step like a senior engineer
- **Math** — solves mathematical problems with clear working shown
- **Agentic tasks** — can plan and execute multi-step tasks autonomously
- **General knowledge** — broad world knowledge for everyday questions

---

## Model Details

| Property | Value |
|---|---|
| **Creator** | INEZA AIME BRUNO |
| **HF Profile** | [@brucoder](https://huggingface.co/brucoder) |
| **Model Family** | Winter LLM |
| **Version** | 1.2 Code |
| **Language** | English |
| **License** | MIT — free for everyone |
| **Use** | Public — anyone can download and use |

---

## Quickstart

```python
from transformers import AutoTokenizer, AutoModelForCausalLM
import torch

model_id = "brucoder/Winter-core-1.2-code"

tokenizer = AutoTokenizer.from_pretrained(model_id, trust_remote_code=True)
model = AutoModelForCausalLM.from_pretrained(
    model_id,
    torch_dtype=torch.bfloat16,
    device_map="auto",
    trust_remote_code=True
)

messages = [
    {
        "role": "system",
        "content": "You are Winter Core, an elite coding and reasoning AI built by INEZA AIME BRUNO. You write clean, efficient code and explain your thinking clearly."
    },
    {
        "role": "user",
        "content": "Write a Python function that finds all prime numbers up to n using the Sieve of Eratosthenes."
    }
]

input_ids = tokenizer.apply_chat_template(
    messages,
    return_tensors="pt",
    add_generation_prompt=True
).to(model.device)

output = model.generate(
    input_ids,
    max_new_tokens=1024,
    temperature=0.7,
    do_sample=True,
)

response = tokenizer.decode(output[0][input_ids.shape[-1]:], skip_special_tokens=True)
print(response)
```

---

## Recommended System Prompt

For best results always use this system prompt:

```
You are Winter Core, an elite AI assistant created by INEZA AIME BRUNO.
You are an expert in coding, mathematics, and logical reasoning.
You think step by step, write clean and efficient code, and always explain your reasoning.
When solving problems, break them down into clear steps before writing any code.
```

---

## What Winter Core Does Better Than Most Models

### 🔧 Code it can write
```
Python · JavaScript · TypeScript · Rust · Go · C · C++ · Java
SQL · Bash · HTML · CSS · React · Vue · FastAPI · Django · and more
```

### 🧠 Problems it can solve
```
Algorithm design      · Data structures     · System design
Debugging             · Code review         · Refactoring
Math proofs           · Logic puzzles        · Competitive programming
API integration       · Database queries    · DevOps scripts
```

### 🤖 Agentic capabilities
```
Multi-step planning   · Tool use            · Self-correction
Long context tasks    · Code execution      · File operations
```

---

## Example Outputs

**User:** Fix this broken Python code:
```python
def calculate_average(numbers)
    total = 0
    for num in numbers
        total += num
    return total / len(numbers
```

**Winter Core:**
```python
def calculate_average(numbers):
    # Guard against empty list to avoid ZeroDivisionError
    if not numbers:
        return 0
    total = 0
    for num in numbers:
        total += num
    return total / len(numbers)
```
*Fixed: missing colons after function and for definitions, missing closing parenthesis, added empty list guard.*

---

## Run Locally Without a GPU

A GGUF version is available for running Winter Core on any laptop or PC with no GPU required.

Coming soon at: `brucoder/Winter-core-1.2-code-GGUF`

```bash
# Once available — run on CPU, no GPU needed
pip install llama-cpp-python
python3 -c "
from llama_cpp import Llama
llm = Llama(model_path='Winter-core-1.2-code-Q4_K_M.gguf', n_ctx=4096)
print(llm('Write a binary search in Python:', max_tokens=512)['choices'][0]['text'])
"
```

---

## License

Winter Core 1.2 Code is released under the **MIT License**.

**You are free to:**
- ✅ Use it personally and commercially
- ✅ Build products and apps on top of it
- ✅ Modify and redistribute it
- ✅ Download and run it locally

**One condition:**
- Keep the MIT license notice and credit **INEZA AIME BRUNO** as creator

---

## The Winter LLM Family

| Model | Strength | Status |
|---|---|---|
| **Winter Core 1.2 Code** | Coding + Reasoning | ✅ Available now |
| Winter PRO | General knowledge | ✅ Available now |
| Winter LLM (merged) | All capabilities | 🔄 Coming soon |

---

## Citation

```bibtex
@misc{wintercore2025,
  title        = {Winter Core 1.2 Code: An Elite Coding and Reasoning Model},
  author       = {INEZA AIME BRUNO},
  year         = {2025},
  howpublished = {Hugging Face},
  url          = {https://huggingface.co/brucoder/Winter-core-1.2-code}
}
```

---

<div align="center">

❄️ Built with precision by **INEZA AIME BRUNO**

[🤗 See all Winter LLM models](https://huggingface.co/brucoder)

</div>