SA

Sarvam 1

SarvamGGUF by bartowskiUpdated 2024-10-29
Download on Hugging Face
Parameters2.53B
Context8K
Layers28
Architecturellama
Smallest quant1.2 GB
Quick answer

Sarvam 1 needs about 3.0 GB of memory at Q4_K_M (a 1.5 GB file) with an 8K context window, so it fits on a 4 GB GPU. This repository has 18 quants, from 1.2 GB (IQ3_M) to 5.1 GB (F16).

Sarvam 1 – 2.53B parameter model optimized for Indian languages

Sarvam 1 is a 2.53B parameter language model built for Indian languages. It supports 10 Indic languages and is trained on a curated corpus of ~4 trillion tokens. The model is a text‑completion model and can be fine‑tuned for downstream tasks.

Good at

  • Optimized for 10 Indian languages
  • High token efficiency
  • Fast inference on modest hardware

Limits

  • No chat or instruction‑following capability
  • No vision support
  • License is non‑commercial

Best for

  • Fine‑tuning for Indic language tasks
  • Text completion in Indian languages
  • Rapid prototyping on limited GPU memory

Who it’s for. Developers and researchers working on Indian language NLP who need a compact, efficient model for fine‑tuning. It is suitable for environments with limited GPU VRAM and for projects that require a non‑commercial license.

Quants & memory

QuantFileEst. memoryvs VRAMResultDownload
IQ3_M1.2 GB 2.6 GB
Fits Get file →
Q3_K_L1.4 GB 2.8 GB
Fits Get file →
IQ4_XS1.4 GB 2.8 GB
Fits Get file →
Q4_01.5 GB 2.9 GB
Fits Get file →
Q4_01.5 GB 2.9 GB
Fits Get file →
Q4_01.5 GB 2.9 GB
Fits Get file →
Q4_01.5 GB 2.9 GB
Fits Get file →
Q4_K_S1.5 GB 2.9 GB
Fits Get file →
Q3_K_XL1.5 GB 2.9 GB
Fits Get file →
Q4_K_M1.5 GB 3.0 GB
Fits Get file →
Q4_K_L1.7 GB 3.1 GB
Fits Get file →
Q5_K_S1.8 GB 3.2 GB
Fits Get file →
Q5_K_M1.8 GB 3.2 GB
Fits Get file →
Q5_K_L1.9 GB 3.3 GB
Fits Get file →
Q6_K2.1 GB 3.5 GB
Fits Get file →
Q6_K_L2.1 GB 3.6 GB
Fits Get file →
Q8_02.7 GB 4.1 GB
Fits Get file →
F165.1 GB 6.5 GB
Fits Get file →

For 8 GB VRAM, use Q4_K_M (file size 1.55 GB, memory 3 GB). For 12 GB VRAM, Q4_K_L (1.65 GB, 3.1 GB). If you have 16 GB VRAM, Q5_K_L (1.89 GB, 3.3 GB) gives higher quality while staying within memory limits. Estimates are approximate.

What GPU do you need for Sarvam 1?

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 Q6_K_L (2.1 GB file, about 3.6 GB in use).
  • 6 GB VRAM: the best quant that fits is Q8_0 (2.7 GB file, about 4.1 GB in use).
  • 8 GB VRAM: the best quant that fits is F16 (5.1 GB file, about 6.5 GB in use).

Which Sarvam 1 quant should you choose?

Q4_K_M is the usual starting point for Sarvam 1: it balances file size and quality well. Here it is a 1.5 GB file.

The Q8_0 file is 2.7 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 IQ3_M at 1.2 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

Architecturellama
Parameters2.53B
Layers28
Trained context8,192 tokens
KV cache (fp16)112 KB per token

Source

Quantized bybartowski
FamilySarvam
Downloads (HF)1,254
Last updated2024-10-29

Frequently asked questions about Sarvam 1

How much VRAM does Sarvam 1 need?

About 3.0 GB at Q4_K_M with an 8K context window. That is the 1.5 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 Sarvam 1 on an 8 GB GPU?

Yes, with F16: the 5.1 GB file needs about 6.5 GB at 8K context. Larger quants in this repository do not fit in 8 GB.

What is the file size of Sarvam 1 GGUF?

It depends on the quant: IQ3_M is 1.2 GB, Q3_K_L is 1.4 GB, IQ4_XS is 1.4 GB, Q4_0 is 1.5 GB, Q4_0 is 1.5 GB, Q4_0 is 1.5 GB and 12 more.

What context length does Sarvam 1 support?

The GGUF header lists a trained context length of 8,192 tokens (8K). The KV cache grows with the context you use: about 0.9 GB at 8K and 0.9 GB at 8K.

Where can I download Sarvam 1 GGUF?

The files are on Hugging Face in the bartowski/sarvam-1-GGUF repository. The download buttons on this page open the file on Hugging Face. ggufmodels.com does not host any model files.

Can I run Sarvam 1 on 8 GB VRAM?

Yes, a Q4_K_M quant (1.55 GB file, ~3 GB memory) fits in 8 GB VRAM with headroom for context and overhead.

Does Sarvam 1 support chat or instruction following?

No, it is a text‑completion model and must be fine‑tuned for chat or instruction tasks.

What license does Sarvam 1 use?

It uses a non‑commercial license; see the LICENSE file for details.

Other GGUFs of sarvam-1

Similar models

Browse all →

More from bartowski

All by bartowski →

More Sarvam models

All Sarvam →

Other tiny models

All tiny models →