File size: 2,325 Bytes
b42a5f2
 
7ea5cfe
b42a5f2
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
---
license: apache-2.0
base_model: mistralai/Mistral-7B-Instruct-v0.2
tags:
- biology
- medical
- mcq
- gguf
- ollama
- mistral
- education
language:
- en
metrics:
- accuracy
---

# Mistral-Biology-MCQ-Strict

This model is a high-precision, zero-hallucination configuration of a Mistral-7B fine-tune, specifically engineered for **Biology Multiple Choice Questions (MCQs)**. It is optimized to provide deterministic, single-letter outputs (A, B, C, D) for automated grading and assessment.

## Model Details

### Model Description

- **Developed by:** HangTuahMalayWarrior (Fine-tuning)
- **Model type:** GGUF (Quantized for Ollama)
- **Language(s) (NLP):** English
- **License:** Apache-2.0
- **Finetuned from model:** Mistral-7B-v0.1 (via QLoRA)

## Uses

### Direct Use

This model is intended for **strictly formatted MCQ processing**. It is designed to be used with a specific Ollama Modelfile that sets the temperature to 0 and utilizes stop sequences to prevent "model bleed" or conversational rambling.

### Out-of-Scope Use

- General purpose chatting or creative writing.
- Non-biological scientific domains (performance may vary).
- Medical advice (This is an educational tool, not a clinical diagnostic tool).

## Bias, Risks, and Limitations

The model has been fine-tuned on a specific subset of Biology MCQs. While it is highly accurate for that domain, it may exhibit "catastrophic forgetting" regarding general knowledge topics not related to biology. 

### Recommendations

To maintain a **100% accuracy target**, users must use the provided Modelfile settings. High temperature settings will cause the model to leak training tags (e.g., `<|start_of_role|>`).

## How to Get Started with the Model

To run this model on Windows using Ollama without hallucinations, create a file named `Modelfile` with the following content:

```dockerfile
FROM hf.co/HangTuahMalayWarrior/mistral-biology-mcq-qlora_gguf:Q4_K_M

# Strict Precision Parameters
PARAMETER temperature 0.0
PARAMETER top_k 1
PARAMETER top_p 0.01

# Stop Sequences to prevent training-tag leaks
PARAMETER stop "<|"
PARAMETER stop "\n"
PARAMETER stop "user"

SYSTEM """You are a strict MCQ answering bot. 
Your output must ONLY be the letter of the correct option (e.g., A, B, C, or D). 
Do not provide explanations. Do not repeat the question."""