UDGAM LABS · Language model, GGUF files · 384M parameters
Gargi-GGUF
GGUF builds of Gargi, a 384M-parameter English model trained from scratch by Udgam Labs, for llama.cpp, Ollama, LM Studio and other GGUF runtimes. See the main model card for evaluation, safety and limitations.
| File | Size | Notes |
|---|---|---|
Gargi-Q8_0.gguf |
410 MB | recommended |
Gargi-Q4_K_M.gguf |
238 MB | for tight memory |
Gargi-f16.gguf |
770 MB | reference precision |
Run
ollama run hf.co/udgamlabs/Gargi-GGUF:Q8_0
llama-cli -hf udgamlabs/Gargi-GGUF:Q8_0 -c 16384
In Ollama, /set think turns on step-by-step thinking (/set nothink turns it off); Gargi writes thoughts mainly on maths problems. Keep
the context at or below ~16K tokens for reliable recall. A ready-made Ollama Modelfile is included.
Quantization
GSM8K with thinking, first 200 test problems, greedy, via Ollama:
| Build | Size | GSM8K (200) | Same answers as f16 |
|---|---|---|---|
| f16 | 770 MB | 39.0 | – |
| Q8_0 (recommended) | 410 MB | 37.5 | 96.5% |
| Q4_K_M | 238 MB | 32.0 | 87% |
Q8_0 matches f16 closely; Q4_K_M loses a few points (each score is ±3.4 points at this sample size). Use Q8_0 unless memory is tight.
Where it falls short
These files share Gargi's limits: see where it falls short on the main model card.
Licence
Apache-2.0.
About Udgam Labs. Udgam Labs is an independent AI research lab. We work from first principles: before we build, we understand why things work, whether we are training a model, fine-tuning one or applying AI to a new problem. Udgam (उद्गम) is Sanskrit for origin. udgamlabs.com