manaca 1b base needs about 2.0 GB of memory at Q4_K_M (a 1.1 GB file) with an 8K context window, so it fits on a 4 GB GPU. This repository has 30 quants, from 0.3 GB (Q1_0) to 3.4 GB (BF16).
1.7B‑parameter Brazilian Portuguese Llama‑style base model
Manacá‑1B is a 1.7‑billion‑parameter decoder‑only LLM trained from scratch for Brazilian Portuguese. It is a base model, pre‑trained without instruction tuning or alignment. The model uses a 64k SentencePiece unigram tokenizer and a 4096‑token context window.
Good at
- Brazilian Portuguese text generation
- Large context handling up to 4096 tokens
- Low‑memory inference with Q4_K_M quantization
Limits
- No instruction tuning or alignment
- No vision support
- Requires 8‑k context for memory estimates
Best for
- Generating Portuguese prose
- Fine‑tuning for domain‑specific Portuguese tasks
- Running on low‑VRAM devices with Q4_K_M
Who it’s for. Developers and researchers needing a reproducible Brazilian Portuguese base model for fine‑tuning or experimentation. The model is open‑licensed (cc‑by‑4.0) and fully reproducible via the provided GitHub pipeline.
Quants & memory
| Quant | File | Est. memory | vs VRAM | Result | |
|---|---|---|---|---|---|
| Q1_0 | 0.3 GB | 1.0 GB | Fits | Get file → | |
| IQ1_S | 0.4 GB | 1.1 GB | Fits | Get file → | |
| IQ1_M | 0.5 GB | 1.2 GB | Fits | Get file → | |
| IQ2_XXS | 0.5 GB | 1.2 GB | Fits | Get file → | |
| IQ2_XS | 0.6 GB | 1.3 GB | Fits | Get file → | |
| Q2_0 | 0.6 GB | 1.3 GB | Fits | Get file → | |
| IQ2_S | 0.6 GB | 1.3 GB | Fits | Get file → | |
| IQ2_M | 0.6 GB | 1.3 GB | Fits | Get file → | |
| Q2_K_S | 0.7 GB | 1.4 GB | Fits | Get file → | |
| Q2_K | 0.7 GB | 1.4 GB | Fits | Get file → | |
| IQ3_XXS | 0.7 GB | 1.4 GB | Fits | Get file → | |
| IQ3_XS | 0.8 GB | 1.5 GB | Fits | Get file → | |
| Q3_K_S | 0.8 GB | 1.5 GB | Fits | Get file → | |
| IQ3_S | 0.8 GB | 1.5 GB | Fits | Get file → | |
| IQ3_M | 0.8 GB | 1.5 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 → | |
| IQ4_XS | 1.0 GB | 1.7 GB | Fits | Get file → | |
| Q4_0 | 1.0 GB | 1.7 GB | Fits | Get file → | |
| IQ4_NL | 1.0 GB | 1.7 GB | Fits | Get file → | |
| Q4_K_S | 1.0 GB | 1.7 GB | Fits | Get file → | |
| Q4_K_M | 1.1 GB | 1.8 GB | Fits | Get file → | |
| Q4_1 | 1.1 GB | 1.8 GB | Fits | Get file → | |
| Q5_0 | 1.2 GB | 1.9 GB | Fits | Get file → | |
| Q5_K_S | 1.2 GB | 1.9 GB | Fits | Get file → | |
| Q5_K_M | 1.2 GB | 1.9 GB | Fits | Get file → | |
| Q5_1 | 1.3 GB | 2.0 GB | Fits | Get file → | |
| Q6_K | 1.4 GB | 2.1 GB | Fits | Get file → | |
| Q8_0 | 1.8 GB | 2.5 GB | Fits | Get file → | |
| BF16 | 3.4 GB | 4.2 GB | Fits | Get file → |
For 8‑GB VRAM, Q4_K_M (0.87 GB) or Q4_K_S (0.65 GB) are suitable. For 4‑GB VRAM, IQ3_M (0.82 GB) or IQ2_M (0.64 GB) fit. Larger models like Q6_K (1.42 GB) or Q8_0 (1.83 GB) need 12‑GB or more VRAM. Memory estimates are approximate.
What GPU do you need for manaca 1b base?
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 Q8_0 (1.8 GB file, about 2.7 GB in use).
- 6 GB VRAM: the best quant that fits is BF16 (3.4 GB file, about 4.4 GB in use).
Which manaca 1b base quant should you choose?
Q4_K_M is the usual starting point for manaca 1b base: it balances file size and quality well. Here it is a 1.1 GB file.
The Q8_0 file is 1.8 GB, about 1.7 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 Q1_0 at 0.3 GB, with the quality trade-offs that come with fewer bits. Lower bit counts mean smaller files and more quality loss.
New to quant names? Read GGUF quantization explained.
Specifications
Architecture from the GGUF header
Source
Frequently asked questions about manaca 1b base
How much VRAM does manaca 1b base need?
About 2.0 GB at Q4_K_M with an 8K context window. That is the 1.1 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 manaca 1b base on an 8 GB GPU?
Yes, with BF16: the 3.4 GB file needs about 4.4 GB at 8K context. Larger quants in this repository do not fit in 8 GB.
What is the file size of manaca 1b base GGUF?
It depends on the quant: Q1_0 is 0.3 GB, IQ1_S is 0.4 GB, IQ1_M is 0.5 GB, IQ2_XXS is 0.5 GB, IQ2_XS is 0.6 GB, Q2_0 is 0.6 GB and 24 more.
What context length does manaca 1b base support?
The GGUF header lists a trained context length of 4,096 tokens (4K). The KV cache grows with the context you use: about 0.4 GB at 8K and 0.2 GB at 4K.
Where can I download manaca 1b base GGUF?
The files are on Hugging Face in the NANI-Nithin/manaca-1b-base-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 manaca 1b base use?
Hugging Face lists the license as cc-by-4.0. 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 Manacá‑1B on 8 GB VRAM?
Yes, using the Q4_K_M or Q4_K_S quantization, which require about 1.8 GB of VRAM for the model and KV cache.
Does the model support English?
The model is trained for Brazilian Portuguese and does not include instruction tuning for English.
Is the model open‑source?
Yes, it is released under cc‑by‑4.0 and the training pipeline is publicly available on GitHub.