106 lines
3.8 KiB
Markdown
106 lines
3.8 KiB
Markdown
---
|
|
license: apache-2.0
|
|
language:
|
|
- en
|
|
base_model:
|
|
- Qwen/Qwen3-1.7B
|
|
pipeline_tag: text-generation
|
|
library_name: transformers
|
|
tags:
|
|
- trl
|
|
- text-generation-inference
|
|
- math
|
|
- code
|
|
---
|
|
|
|
|
|
|
|
# **Mintaka-Qwen3-1.6B-V3.1**
|
|
|
|
> Mintaka-Qwen3-1.6B-V3.1 is a high-efficiency, science-focused reasoning model **based on Qwen-1.6B** and trained on **DeepSeek v3.1 synthetic traces (10,000 entries)**. It is optimized for random event simulation, logical-problem analysis, and structured scientific reasoning. The model balances symbolic precision with lightweight deployment, making it suitable for researchers, educators, and developers seeking efficient reasoning under constrained compute.
|
|
|
|
|
|
> \[!note]
|
|
> GGUF: https://huggingface.co/prithivMLmods/Mintaka-Qwen3-1.6B-V3.1-GGUF
|
|
|
|
---
|
|
|
|
## **Key Features**
|
|
|
|
1. **Scientific Reasoning & Chain-of-Thought**
|
|
Trained on **10,000 synthetic traces** from the **DeepSeek v3.1** dataset, designed to enhance step-by-step analytical and probabilistic reasoning for simulation tasks and logical puzzles.
|
|
|
|
2. **Advanced Code Reasoning & Generation**
|
|
Supports multi-language coding with explanations, optimization hints, and error detection—useful for algorithm synthesis, debugging, and prototyping.
|
|
|
|
3. **Random Event Simulation & Logical Analysis**
|
|
Tailored for stochastic event simulations, scenario analysis, and formal logical problem solving.
|
|
|
|
4. **Hybrid Symbolic-AI Thinking**
|
|
Combines structured logic, chain-of-thought reasoning, and open-ended inference to deliver robust performance on STEM and simulation tasks.
|
|
|
|
5. **Structured Output Mastery**
|
|
Generates output in **LaTeX**, **Markdown**, **JSON**, **CSV**, and **YAML**, suited for technical documentation, experiments, and dataset generation.
|
|
|
|
6. **Optimized Lightweight Footprint for Versatile Deployment**
|
|
Balances performance and efficiency — deployable on **mid-range GPUs**, **offline clusters**, and **edge AI systems**.
|
|
|
|
---
|
|
|
|
## **Quickstart with Transformers**
|
|
|
|
```python
|
|
from transformers import AutoModelForCausalLM, AutoTokenizer
|
|
|
|
model_name = "prithivMLmods/Mintaka-Qwen3-1.6B-V3.1"
|
|
|
|
model = AutoModelForCausalLM.from_pretrained(
|
|
model_name,
|
|
torch_dtype="auto",
|
|
device_map="auto"
|
|
)
|
|
tokenizer = AutoTokenizer.from_pretrained(model_name)
|
|
|
|
prompt = "Explain the difference between deterministic simulation and stochastic simulation with examples."
|
|
|
|
messages = [
|
|
{"role": "system", "content": "You are a scientific tutor skilled in reasoning, simulation design, and logical analysis."},
|
|
{"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]
|
|
print(response)
|
|
```
|
|
|
|
---
|
|
|
|
## **Intended Use**
|
|
|
|
* Random event simulation, scenario analysis, and probabilistic reasoning
|
|
* Logical-problem analysis and structured scientific tutoring
|
|
* Research assistant for physics, computational biology, and interdisciplinary simulation domains
|
|
* Structured technical data and experiment result generation
|
|
* Deployment in mid-resource environments requiring efficient reasoning
|
|
|
|
## **Limitations**
|
|
|
|
* Not tuned for long-form creative writing or conversational small talk
|
|
* Context window limitations may hinder multi-document or full codebase analysis
|
|
* Optimized specifically for simulation and logical analysis tasks—general chat may underperform
|
|
* Prioritizes structured logic and reproducibility over emotional tone |