GR

Granite 20b Functioncalling

GraniteGGUF by RichardErkhovUpdated 2024-09-17
Download on Hugging Face
Parameters20B
Context8K
Layers52
Architecturestarcoder
Smallest quant7.9 GB
Quick answer

Granite 20b Functioncalling needs about 13.5 GB of memory at Q4_K_M (a 12.8 GB file) with an 8K context window, so it fits on a 16 GB GPU. This repository has 22 quants, from 7.9 GB (Q2_K) to 21.5 GB (Q8_0).

Granite 20b Functioncalling – a 20B starcoder model with function‑calling finetuning

Granite 20b Functioncalling is a 20‑billion‑parameter starcoder model fine‑tuned for function‑calling tasks. It supports a 8192‑token context window and is released under the Apache‑2.0 license. The model is available in multiple GGUF quantizations.

Good at

  • Function‑calling instruction handling
  • Multi‑task function chaining
  • Nested function detection

Limits

  • No vision support
  • Requires 8‑k context for full capability
  • License is Apache‑2.0

Best for

  • Developers building function‑calling APIs
  • Code generation with function prompts
  • Automated code documentation

Who it’s for. The model is suitable for developers and researchers who need a function‑calling capable LLM for code or API interactions. It is ideal for those who can run GGUF models locally and have sufficient GPU memory.

Quants & memory

QuantFileEst. memoryvs VRAMResultDownload
Q2_K7.9 GB 8.6 GB
Too big Get file →
IQ3_XS8.7 GB 9.4 GB
Too big Get file →
IQ3_S8.9 GB 9.7 GB
Too big Get file →
Q3_K_S8.9 GB 9.7 GB
Too big Get file →
IQ3_M9.6 GB 10.3 GB
Too big Get file →
Q3_K10.6 GB 11.3 GB
Too big Get file →
Q3_K_M10.6 GB 11.3 GB
Too big Get file →
IQ4_XS11.1 GB 11.8 GB
Too big Get file →
Q4_011.6 GB 12.3 GB
Too big Get file →
IQ4_NL11.7 GB 12.4 GB
Too big Get file →
Q4_K_S11.7 GB 12.4 GB
Too big Get file →
Q3_K_L11.7 GB 12.5 GB
Too big Get file →
Q4_112.8 GB 13.5 GB
Too big Get file →
Q4_K12.8 GB 13.5 GB
Too big Get file →
Q4_K_M12.8 GB 13.5 GB
Too big Get file →
Q5_014.0 GB 14.7 GB
Too big Get file →
Q5_K_S14.0 GB 14.7 GB
Too big Get file →
Q5_K14.8 GB 15.5 GB
Too big Get file →
Q5_K_M14.8 GB 15.5 GB
Too big Get file →
Q5_115.2 GB 16.0 GB
Too big Get file →
Q6_K16.6 GB 17.4 GB
Too big Get file →
Q8_021.5 GB 22.2 GB
Too big Get file →

For 8‑GB VRAM, use Q2_K (7.93 GB, ~8.6 GB RAM). For 12‑GB VRAM, Q4_0 (11.55 GB, ~12.3 GB RAM) is a good balance. For 16‑GB VRAM, Q5_0 (14.02 GB, ~14.7 GB RAM) provides higher quality. Estimates are approximate.

What GPU do you need for Granite 20b Functioncalling?

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: no quant fits at an 8K context.
  • 6 GB VRAM: no quant fits at an 8K context.
  • 8 GB VRAM: no quant fits at an 8K context.
  • 12 GB VRAM: the best quant that fits is IQ4_XS (11.1 GB file, about 11.8 GB in use).
  • 16 GB VRAM: the best quant that fits is Q5_1 (15.2 GB file, about 16.0 GB in use).
  • 24 GB VRAM: the best quant that fits is Q8_0 (21.5 GB file, about 22.2 GB in use).

Which Granite 20b Functioncalling quant should you choose?

Q4_K_M is the usual starting point for Granite 20b Functioncalling: it balances file size and quality well. Here it is a 12.8 GB file.

The Q8_0 file is 21.5 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 Q2_K at 7.9 GB, with the quality trade-offs that come with fewer bits. Use it only when memory is very tight. Expect a clear drop in quality, which is usually worse on small models and easier to live with on large ones.

New to quant names? Read GGUF quantization explained.

Specifications

Architecture from the GGUF header

Architecturestarcoder
Parameters20B
Layers52
Trained context8,192 tokens
KV cache (fp16)26 KB per token

Source

Quantized byRichardErkhov
FamilyGranite
Licenseapache-2.0
Downloads (HF)4,164
Last updated2024-09-17

Frequently asked questions about Granite 20b Functioncalling

How much VRAM does Granite 20b Functioncalling need?

About 13.5 GB at Q4_K_M with an 8K context window. That is the 12.8 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 Granite 20b Functioncalling on an 8 GB GPU?

Not at an 8K context. The smallest file needs about 8.6 GB, so you would need a larger GPU or CPU offloading.

What is the file size of Granite 20b Functioncalling GGUF?

It depends on the quant: Q2_K is 7.9 GB, IQ3_XS is 8.7 GB, IQ3_S is 8.9 GB, Q3_K_S is 8.9 GB, IQ3_M is 9.6 GB, Q3_K is 10.6 GB and 16 more.

What context length does Granite 20b Functioncalling support?

The GGUF header lists a trained context length of 8,192 tokens (8K). The KV cache grows with the context you use: about 0.2 GB at 8K and 0.2 GB at 8K.

Where can I download Granite 20b Functioncalling GGUF?

The files are on Hugging Face in the RichardErkhov/ibm-granite_-_granite-20b-functioncalling-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 Granite 20b Functioncalling 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 Granite 20b Functioncalling on 8 GB VRAM?

Yes, the Q2_K quantization is about 7.93 GB and requires roughly 8.6 GB of RAM at 8‑k context.

Does this model support image inputs?

No, the model has no vision capabilities.

What license applies to this model?

The model is released under the Apache‑2.0 license.

Similar models

Browse all →

More from RichardErkhov

All by RichardErkhov →

More Granite models

All Granite →

Other medium models

All medium models →