VI

it-5.3-fp16-32k

VikhrmodelsGGUF by VikhrmodelsUpdated 2024-06-13
Download on Hugging Face
Parameters7.63B
Context32K
Layers32
Architecturellama
Smallest quant1.8 GB
Quick answer

it-5.3-fp16-32k needs about 6.2 GB of memory at Q4_K_M (a 4.6 GB file) with an 8K context window, so it fits on a 8 GB GPU. This repository has 31 quants, from 1.8 GB (IQ1_S) to 15.3 GB (F16). Longer context needs more: at 32K tokens the KV cache alone adds about 4.3 GB.

it-5.3-fp16-32k is a 7.63B-parameter Vikhrmodels model in GGUF format, quantized by Vikhrmodels. Its trained context length is 32K tokens. The repository has 31 quant files, from 1.8 GB (IQ1_S) to 15.3 GB (F16).

Quants & memory

QuantFileEst. memoryvs VRAMResultDownload
IQ1_S1.8 GB 3.4 GB
Fits Get file →
IQ1_M2.0 GB 3.5 GB
Fits Get file →
IQ2_XXS2.2 GB 3.8 GB
Fits Get file →
IQ2_XS2.4 GB 4.0 GB
Fits Get file →
IQ2_S2.5 GB 4.1 GB
Fits Get file →
IQ2_M2.7 GB 4.3 GB
Fits Get file →
Q2_K_S2.8 GB 4.3 GB
Fits Get file →
Q2_K2.9 GB 4.5 GB
Fits Get file →
IQ3_XXS3.0 GB 4.6 GB
Fits Get file →
IQ3_XS3.3 GB 4.8 GB
Fits Get file →
Q3_K_S3.4 GB 5.0 GB
Fits Get file →
IQ3_S3.4 GB 5.0 GB
Fits Get file →
IQ3_M3.5 GB 5.1 GB
Fits Get file →
Q3_K3.8 GB 5.3 GB
Fits Get file →
Q3_K_M3.8 GB 5.3 GB
Fits Get file →
Q3_K_L4.1 GB 5.6 GB
Fits Get file →
IQ4_XS4.2 GB 5.7 GB
Fits Get file →
Q4_04.4 GB 6.0 GB
Fits Get file →
IQ4_NL4.4 GB 6.0 GB
Fits Get file →
Q4_K_S4.4 GB 6.0 GB
Fits Get file →
Q4_K4.6 GB 6.2 GB
Fits Get file →
Q4_K_M4.6 GB 6.2 GB
Fits Get file →
Q4_14.8 GB 6.4 GB
Fits Get file →
Q5_05.3 GB 6.9 GB
Fits Get file →
Q5_K_S5.3 GB 6.9 GB
Fits Get file →
Q5_K5.4 GB 7.0 GB
Fits Get file →
Q5_K_M5.4 GB 7.0 GB
Fits Get file →
Q5_15.7 GB 7.3 GB
Tight Get file →
Q6_K6.3 GB 7.8 GB
Tight Get file →
Q8_08.1 GB 9.7 GB
Too big Get file →
F1615.3 GB 16.8 GB
Too big Get file →

What GPU do you need for it-5.3-fp16-32k?

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 IQ2_XS (2.4 GB file, about 4.0 GB in use).
  • 6 GB VRAM: the best quant that fits is Q4_K_S (4.4 GB file, about 6.0 GB in use).
  • 8 GB VRAM: the best quant that fits is Q6_K (6.3 GB file, about 7.8 GB in use).
  • 12 GB VRAM: the best quant that fits is Q8_0 (8.1 GB file, about 9.7 GB in use).
  • 16 GB VRAM: the best quant that fits is Q8_0 (8.1 GB file, about 9.7 GB in use).
  • 24 GB VRAM: the best quant that fits is F16 (15.3 GB file, about 16.8 GB in use).

Which it-5.3-fp16-32k quant should you choose?

Q4_K_M is the usual starting point for it-5.3-fp16-32k: it balances file size and quality well. Here it is a 4.6 GB file.

The Q8_0 file is 8.1 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 IQ1_S at 1.8 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

Architecturellama
Parameters7.63B
Layers32
Trained context32,768 tokens
KV cache (fp16)128 KB per token

Source

Quantized byVikhrmodels
Licenseapache-2.0
Downloads (HF)319
Last updated2024-06-13

Frequently asked questions about it-5.3-fp16-32k

How much VRAM does it-5.3-fp16-32k need?

About 6.2 GB at Q4_K_M with an 8K context window. That is the 4.6 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 it-5.3-fp16-32k on an 8 GB GPU?

Yes, with Q6_K: the 6.3 GB file needs about 7.8 GB at 8K context. Larger quants in this repository do not fit in 8 GB.

What is the file size of it-5.3-fp16-32k GGUF?

It depends on the quant: IQ1_S is 1.8 GB, IQ1_M is 2.0 GB, IQ2_XXS is 2.2 GB, IQ2_XS is 2.4 GB, IQ2_S is 2.5 GB, IQ2_M is 2.7 GB and 25 more.

What context length does it-5.3-fp16-32k support?

The GGUF header lists a trained context length of 32,768 tokens (32K). The KV cache grows with the context you use: about 1.1 GB at 8K and 4.3 GB at 32K.

Where can I download it-5.3-fp16-32k GGUF?

The files are on Hugging Face in the Vikhrmodels/it-5.3-fp16-32k-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 it-5.3-fp16-32k 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.

Similar models

Browse all →

More from Vikhrmodels

All by Vikhrmodels →

More Vikhrmodels models

All Vikhrmodels →

Other small models

All small models →