File size: 1,821 Bytes
f27adea 19ccce6 f27adea 19ccce6 f27adea 19ccce6 f27adea 19ccce6 f27adea 19ccce6 f27adea |
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 72 73 74 |
---
language:
- en
license: apache-2.0
tags:
- text-generation
- gguf
- gemma
- summarization
base_model: google/gemma-3-270m-it
model_type: gemma
---
# catnip-summarizer
Fine-tuned Gemma-3-270M for task summarization and branch naming. It's meant to be used summarize user requests to coding agents. For example:
```shell
ollama run --verbose hf.co/vanpelt/catnip-summarizer "Help me make the vibes good"
Make Good Vibes
feat/make-good-vibes
```
The summary and git branch name can be used by tools like [Catnip](https://github.com/wandb/catnip) to keep multiple coding agent sessions organized.
## Model Details
- **Base Model**: google/gemma-3-270m-it
- **Format**: GGUF
- **Quantization**: Q4_K_M
- **Use Case**: Generating concise task titles and git branch names
## Training
- **Training Run**: [https://wandb.ai/vanpelt/summarizer/runs/jh27ahps](https://wandb.ai/vanpelt/summarizer/runs/jh27ahps)
- **Repo**: [https://github.com/vanpelt/summarizer](https://github.com/vanpelt/summarizer)
## Usage
### Web Demo
You can [try the model](https://vanpelt.github.io/summarizer/) in your browser!
### With Ollama
```bash
ollama run hf.co/vanpelt/catnip-summarizer "Help me make the vibes good"
```
### With llama.cpp
```bash
# Download the GGUF file
huggingface-cli download vanpelt/catnip-summarizer gemma3-270m-summarizer-Q4_K_M.gguf
# Run with llama.cpp
./main -m gemma3-270m-summarizer-Q4_K_M.gguf -p 'Your prompt here'
```
## Files
- `tokenizer.json` (31.8 MB)
- `tokenizer_config.json` (1.1 MB)
- `added_tokens.json` (0.0 MB)
- `chat_template.jinja` (0.0 MB)
- `Modelfile` (0.0 MB)
- `template` (0.0 MB)
- `system` (0.0 MB)
- `gemma3-270m-summarizer-Q4_K_M.gguf` (241.4 MB)
- `special_tokens_map.json` (0.0 MB)
- `config.json` (0.0 MB)
- `params` (0.0 MB)
- `tokenizer.model` (4.5 MB)
|