GA

gaggle-nanochat-pretrained-climbmix-20260924

GaggleGGUF by castoriniUpdated 2026-10-17
Download on Hugging Face
Parameters3.3B
Context2K
Layers34
Architecturegaggle
Smallest quant2.6 GB
Quick answer

gaggle-nanochat-pretrained-climbmix-20260924 needs about 5.5 GB of memory at Q4_K_M (a 2.6 GB file) with an 8K context window, so it fits on a 6 GB GPU. This repository has 3 quants, from 2.6 GB (Q4_K_M) to 6.6 GB (BF16).

3.3B‑parameter Gaggle decoder‑only model

gaggle‑nanochat‑pretrained‑climbmix‑20260924 is a 3.3B‑parameter decoder‑only language model. It was trained from scratch on one pass over ClimbMix (287B tokens) using nanochat. The model continues text without instruction tuning or RLHF.

Good at

  • Text continuation
  • Coding tasks
  • Large‑context generation

Limits

  • No instruction following
  • No vision support
  • Limited to 2048‑token context

Best for

  • Generating code snippets
  • Auto‑completion in IDEs
  • Chatbot text generation

Who it’s for. Developers and researchers who need a lightweight, open‑source language model for code generation and text continuation. The model is suitable for environments where instruction tuning is not required.

Quants & memory

QuantFileEst. memoryvs VRAMResultDownload
Q4_K_M2.6 GB 3.7 GB
Fits Get file →
Q8_03.5 GB 4.6 GB
Fits Get file →
BF166.6 GB 7.7 GB
Tight Get file →

For 8 GB VRAM, use Q4_K_M (2.56 GB file, ~5.5 GB memory at 8k context). For 12 GB VRAM, Q8_0 (3.49 GB file, ~6.4 GB memory). For larger budgets, BF16 (6.57 GB file, ~9.5 GB memory).

What GPU do you need for gaggle-nanochat-pretrained-climbmix-20260924?

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 (2.6 GB file, about 5.5 GB in use).
  • 8 GB VRAM: the best quant that fits is Q8_0 (3.5 GB file, about 6.4 GB in use).
  • 12 GB VRAM: the best quant that fits is BF16 (6.6 GB file, about 9.5 GB in use).

Which gaggle-nanochat-pretrained-climbmix-20260924 quant should you choose?

Q4_K_M is the usual starting point for gaggle-nanochat-pretrained-climbmix-20260924: it balances file size and quality well. Here it is a 2.6 GB file.

The Q8_0 file is 3.5 GB, about 1.4 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.

New to quant names? Read GGUF quantization explained.

Specifications

Architecture from the GGUF header

Architecturegaggle
Parameters3.3B
Layers34
Trained context2,048 tokens
KV cache (fp16)289 KB per token

Source

Quantized bycastorini
FamilyGaggle
Licensemit
Downloads (HF)1,176
Last updated2026-10-17

Frequently asked questions about gaggle-nanochat-pretrained-climbmix-20260924

How much VRAM does gaggle-nanochat-pretrained-climbmix-20260924 need?

About 5.5 GB at Q4_K_M with an 8K context window. That is the 2.6 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 gaggle-nanochat-pretrained-climbmix-20260924 on an 8 GB GPU?

Yes, with Q8_0: the 3.5 GB file needs about 6.4 GB at 8K context. Larger quants in this repository do not fit in 8 GB.

What is the file size of gaggle-nanochat-pretrained-climbmix-20260924 GGUF?

It depends on the quant: Q4_K_M is 2.6 GB, Q8_0 is 3.5 GB, BF16 is 6.6 GB.

What context length does gaggle-nanochat-pretrained-climbmix-20260924 support?

The GGUF header lists a trained context length of 2,048 tokens (2K). The KV cache grows with the context you use: about 2.4 GB at 8K and 0.6 GB at 2K.

Where can I download gaggle-nanochat-pretrained-climbmix-20260924 GGUF?

The files are on Hugging Face in the castorini/gaggle-nanochat-pretrained-climbmix-20260924 repository. The download buttons on this page open the file on Hugging Face. ggufmodels.com does not host any model files.

What license does gaggle-nanochat-pretrained-climbmix-20260924 use?

Hugging Face lists the license as mit. 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, the Q4_K_M quantized file is 2.56 GB and requires about 5.5 GB of VRAM at 8k context.

Does this model support instruction following?

No, it has no post‑training instruction tuning or RLHF, so it continues text rather than following instructions.

What context length does it support?

The model was trained with a context of 2048 tokens and is limited to that length.

Similar models

Browse all →

More from castorini

All by castorini →

Other tiny models

All tiny models →