---
tags:
  - gguf
  - qwen2
  - finance
  - news-analysis
  - stock-market
base_model: Qwen/Qwen2.5-0.5B-Instruct
license: apache-2.0
---

# qwen-0.5b-stocks-news-v2 (GGUF)

GGUF quantized versions of [qwen-0.5b-stocks-news-v2](https://huggingface.co/VaibTFU/qwen-0.5b-stocks-news-v2).

## Model Description

Fine-tuned **Qwen2.5-0.5B-Instruct** for financial news analysis. Outputs structured JSON with:
- `impactTerm`: LONGTERM / SHORTTERM
- `impactScope`: GLOBAL / INDUSTRY / STOCK
- `scopeItems`: list of affected stocks/industries
- `highlights`: key information summary
- `sentiment`: POSITIVE / NEGATIVE / NEUTRAL

## Available Quantizations

| Quantization | Size | Quality | Use Case |
|:------------|:-----|:--------|:---------|
| F16 | ~1 GB | Highest | Reference / GPU inference |
| Q8_0 | ~530 MB | Very High | Desktop / server inference |
| Q4_K_M | ~320 MB | Good | Mobile / edge / low-RAM |

## Usage

### With llama.cpp

```bash
./llama-cli -m qwen-0.5b-stocks-news-v2-q4_k_m.gguf \
  -p '<|im_start|>system
You are a financial news AI. Analyze the given news article and output its scope, impact, highlights, and sentiment in JSON format. /no_think<|im_end|>
<|im_start|>user
RBI holds repo rate at 6.5% in MPC meeting<|im_end|>
<|im_start|>assistant
' \
  -n 256 --temp 0.7
```

### With Ollama

Create a `Modelfile`:
```
FROM ./qwen-0.5b-stocks-news-v2-q4_k_m.gguf

TEMPLATE """<|im_start|>system
{{ .System }}<|im_end|>
<|im_start|>user
{{ .Prompt }}<|im_end|>
<|im_start|>assistant
"""

SYSTEM "You are a financial news AI. Analyze the given news article and output its scope, impact, highlights, and sentiment in JSON format. /no_think"

PARAMETER temperature 0.7
PARAMETER top_p 0.8
PARAMETER stop <|im_end|>
```

Then:
```bash
ollama create qwen-0.5b-stocks-news-v2 -f Modelfile
ollama run qwen-0.5b-stocks-news-v2 "Infosys Q3 profit rises 11%"
```