PH

Iceball Phi 3.5 mini instruct shareGPT v1.0 i1

PhiGGUF by mradermacherUpdated 2024-08-23
Download on Hugging Face
Parameters3.82B
Context128K
Layers32
Architecturellama
Smallest quant0.9 GB
Quick answer

Iceball Phi 3.5 mini instruct shareGPT v1.0 i1 needs about 6.0 GB of memory at Q4_K_M (a 2.3 GB file) with an 8K context window, so it fits on a 6 GB GPU. This repository has 21 quants, from 0.9 GB (IQ1_S) to 3.1 GB (Q6_K). Longer context needs more: at 32K tokens the KV cache alone adds about 12.9 GB.

Iceball Phi 3.5 mini instruct shareGPT v1.0 i1 – a 3.82B Phi model in GGUF format

Iceball Phi 3.5 mini instruct shareGPT v1.0 i1 is a 3.82B parameter Phi model quantized for local inference. It supports a 131072 token context length and is released under the apache-2.0 license. The model is available in multiple GGUF quantizations for different memory budgets.

Good at

  • High‑quality instruction following
  • Strong reasoning in code, math and logic
  • Low‑latency inference on modest GPUs

Limits

  • No vision support
  • Requires 8K context for memory estimates
  • Large memory needs for higher‑quality quants

Best for

  • Chat‑based applications
  • Research prototypes on limited hardware
  • Rapid prototyping of instruction‑tuned LLMs

Who it’s for. Developers and researchers who need a Phi‑3.5 mini model that can run locally with a 131072 token context. The model is suitable for commercial or research use under the apache‑2.0 license.

Quants & memory

QuantFileEst. memoryvs VRAMResultDownload
IQ1_S0.9 GB 4.6 GB
Fits Get file →
IQ1_M1.0 GB 4.7 GB
Fits Get file →
IQ2_XXS1.1 GB 4.8 GB
Fits Get file →
IQ2_XS1.2 GB 4.9 GB
Fits Get file →
IQ2_S1.3 GB 5.0 GB
Fits Get file →
IQ2_M1.3 GB 5.1 GB
Fits Get file →
Q2_K1.4 GB 5.2 GB
Fits Get file →
IQ3_XXS1.5 GB 5.2 GB
Fits Get file →
IQ3_XS1.6 GB 5.3 GB
Fits Get file →
IQ3_S1.7 GB 5.4 GB
Fits Get file →
Q3_K_S1.7 GB 5.4 GB
Fits Get file →
IQ3_M1.8 GB 5.5 GB
Fits Get file →
Q3_K_M1.9 GB 5.6 GB
Fits Get file →
Q3_K_L2.0 GB 5.8 GB
Fits Get file →
IQ4_XS2.1 GB 5.8 GB
Fits Get file →
Q4_02.2 GB 5.9 GB
Fits Get file →
Q4_K_S2.2 GB 5.9 GB
Fits Get file →
Q4_K_M2.3 GB 6.0 GB
Fits Get file →
Q5_K_S2.6 GB 6.4 GB
Fits Get file →
Q5_K_M2.7 GB 6.4 GB
Fits Get file →
Q6_K3.1 GB 6.9 GB
Fits Get file →

For 8 GB VRAM, use IQ1_S (0.88 GB) or IQ1_M (0.95 GB). If 12 GB is available, IQ2_S (1.26 GB) or IQ2_M (1.35 GB) provide better quality. For 16 GB or more, consider IQ3_S (1.68 GB) or Q4_K_M (2.32 GB) for higher accuracy. Memory estimates are approximate.

What GPU do you need for Iceball Phi 3.5 mini instruct shareGPT v1.0 i1?

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: the best quant that fits is Q4_K_M (2.3 GB file, about 6.0 GB in use).
  • 8 GB VRAM: the best quant that fits is Q6_K (3.1 GB file, about 6.9 GB in use).

Which Iceball Phi 3.5 mini instruct shareGPT v1.0 i1 quant should you choose?

Q4_K_M is the usual starting point for Iceball Phi 3.5 mini instruct shareGPT v1.0 i1: it balances file size and quality well. Here it is a 2.3 GB file.

If memory is tight, the smallest file is IQ1_S at 0.9 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

Architecturellama
Parameters3.82B
Layers32
Trained context131,072 tokens
KV cache (fp16)384 KB per token

Frequently asked questions about Iceball Phi 3.5 mini instruct shareGPT v1.0 i1

How much VRAM does Iceball Phi 3.5 mini instruct shareGPT v1.0 i1 need?

About 6.0 GB at Q4_K_M with an 8K context window. That is the 2.3 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 Iceball Phi 3.5 mini instruct shareGPT v1.0 i1 on an 8 GB GPU?

Yes, with Q6_K: the 3.1 GB file needs about 6.9 GB at 8K context. Larger quants in this repository do not fit in 8 GB.

What is the file size of Iceball Phi 3.5 mini instruct shareGPT v1.0 i1 GGUF?

It depends on the quant: IQ1_S is 0.9 GB, IQ1_M is 1.0 GB, IQ2_XXS is 1.1 GB, IQ2_XS is 1.2 GB, IQ2_S is 1.3 GB, IQ2_M is 1.3 GB and 15 more.

What context length does Iceball Phi 3.5 mini instruct shareGPT v1.0 i1 support?

The GGUF header lists a trained context length of 131,072 tokens (128K). The KV cache grows with the context you use: about 3.2 GB at 8K and 12.9 GB at 32K.

Where can I download Iceball Phi 3.5 mini instruct shareGPT v1.0 i1 GGUF?

The files are on Hugging Face in the mradermacher/Iceball-Phi-3.5-mini-instruct-shareGPT-v1.0-i1-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 Iceball Phi 3.5 mini instruct shareGPT v1.0 i1 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 this model on 8 GB VRAM?

Yes, use IQ1_S or IQ1_M quantizations. They require about 4.6–4.7 GB of VRAM at 8K context.

Does this model support vision inputs?

No, the model does not include vision capabilities.

What context length does it support?

It supports a 131072 token context length, but memory estimates are based on 8K context.

Other GGUFs of Iceball-Phi-3.5-mini-instruct-shareGPT-v1.0

Similar models

Browse all →

More from mradermacher

All by mradermacher →

More Phi models

All Phi →

Other small models

All small models →