Bangla Gemma Pro needs about 1.7 GB of memory at Q4_K_M (a 1.0 GB file) with an 8K context window, so it fits on a 4 GB GPU. This repository has 12 quants, from 0.9 GB (Q3_K_S) to 2.6 GB (F16). Longer context needs more: at 32K tokens the KV cache alone adds about 0.9 GB.
Bangla Gemma Pro – a QLoRA fine‑tune of Gemma3 on Bangla instruction data
Bangla Gemma Pro is a 1.3B‑parameter Gemma3 model fine‑tuned with QLoRA on Bangla instruction data. The fine‑tune did not produce usable Bangla generation and is published as a documented negative result.
Good at
- QLoRA fine‑tune on Bangla instruction data
- Small model size
- Apache‑2.0 license
Limits
- Bangla generation unreliable
- No vision support
- Not suitable for deployment
Best for
- Researching negative results
- Benchmarking small Bangla models
- Analyzing fine‑tune failures
Who it’s for. Researchers and developers interested in the limits of QLoRA fine‑tuning on Bangla instruction data. The model is not intended for production use.
Quants & memory
| Quant | File | Est. memory | vs VRAM | Result | |
|---|---|---|---|---|---|
| Q3_K_S | 0.9 GB | 1.6 GB | Fits | Get file → | |
| Q2_K | 0.9 GB | 1.6 GB | Fits | Get file → | |
| IQ4_XS | 0.9 GB | 1.6 GB | Fits | Get file → | |
| Q3_K_M | 0.9 GB | 1.6 GB | Fits | Get file → | |
| Q3_K_L | 0.9 GB | 1.6 GB | Fits | Get file → | |
| Q4_K_S | 1.0 GB | 1.7 GB | Fits | Get file → | |
| Q4_K_M | 1.0 GB | 1.7 GB | Fits | Get file → | |
| Q5_K_S | 1.1 GB | 1.8 GB | Fits | Get file → | |
| Q5_K_M | 1.1 GB | 1.8 GB | Fits | Get file → | |
| Q6_K | 1.3 GB | 2.1 GB | Fits | Get file → | |
| Q8_0 | 1.4 GB | 2.1 GB | Fits | Get file → | |
| F16 | 2.6 GB | 3.3 GB | Fits | Get file → |
For 8 GB VRAM, use Q4_K_S (0.99 GB file, ~1.7 GB memory at 8 k context). For 12 GB VRAM, Q5_K_M (1.08 GB file, ~1.8 GB memory) is a good balance. Larger VRAM allows Q6_K or Q8_0 for higher precision.
What GPU do you need for Bangla Gemma Pro?
These figures add the file size, an fp16 KV cache at an 8K context window (worked out from this model’s own header) and about 0.5 GB for the runtime. They are estimates, so leave a little headroom, and expect to use less context or a smaller quant if you are right at the limit.
- 4 GB VRAM: the best quant that fits is F16 (2.6 GB file, about 3.3 GB in use).
Which Bangla Gemma Pro quant should you choose?
Q4_K_M is the usual starting point for Bangla Gemma Pro: it balances file size and quality well. Here it is a 1.0 GB file.
The Q8_0 file is 1.4 GB, about 1.4 times the size of Q4_K_M. It keeps the weights closer to the original, which only matters if you have the memory to spare.
If memory is tight, the smallest file is Q3_K_S at 0.9 GB, with the quality trade-offs that come with fewer bits. A reasonable choice when a 4-bit file will not fit, mostly on larger models. Quality loss is noticeable but often acceptable.
New to quant names? Read GGUF quantization explained.
Specifications
Architecture from the GGUF header
Source
Frequently asked questions about Bangla Gemma Pro
How much VRAM does Bangla Gemma Pro need?
About 1.7 GB at Q4_K_M with an 8K context window. That is the 1.0 GB file, plus the KV cache and roughly 0.5 GB of runtime overhead. Smaller quants need less and longer context needs more.
Can I run Bangla Gemma Pro on an 8 GB GPU?
Yes, with F16: the 2.6 GB file needs about 3.3 GB at 8K context. Larger quants in this repository do not fit in 8 GB.
What is the file size of Bangla Gemma Pro GGUF?
It depends on the quant: Q3_K_S is 0.9 GB, Q2_K is 0.9 GB, IQ4_XS is 0.9 GB, Q3_K_M is 0.9 GB, Q3_K_L is 0.9 GB, Q4_K_S is 1.0 GB and 6 more.
What context length does Bangla Gemma Pro support?
The GGUF header lists a trained context length of 32,768 tokens (32K). The KV cache grows with the context you use: about 0.2 GB at 8K and 0.9 GB at 32K.
Where can I download Bangla Gemma Pro GGUF?
The files are on Hugging Face in the mradermacher/bangla-gemma-pro-GGUF repository. The download buttons on this page open the file on Hugging Face. ggufmodels.com does not host any model files.
What license does Bangla Gemma Pro use?
Hugging Face lists the license as apache-2.0. Licenses can differ between the quantized files and the original model, so check the repository and the base model before commercial use.
Can I run Bangla Gemma Pro on 8 GB VRAM?
Yes, the Q4_K_S quant uses about 1.7 GB of VRAM at 8 k context, which fits in 8 GB.
Does this model generate correct Bangla text?
No, the fine‑tune produces ungrammatical and factually incorrect Bangla; it is a documented negative result.
Is this model suitable for deployment?
No, it is not intended for production use and is provided only for research purposes.