AZ

MN Halide 12b v1.0

AzureGGUF by QuantFactoryUpdated 2024-09-24
Download on Hugging Face
Parameters12.2B
Context1000K
Layers40
Architecturellama
Smallest quant4.8 GB
Quick answer

MN Halide 12b v1.0 needs about 9.3 GB of memory at Q4_K_M (a 7.5 GB file) with an 8K context window, so it fits on a 12 GB GPU. This repository has 14 quants, from 4.8 GB (Q2_K) to 13.0 GB (Q8_0). Longer context needs more: at 32K tokens the KV cache alone adds about 5.4 GB.

MN Halide 12b v1.0 – a 12.2B parameter Llama model focused on reasoning

MN Halide 12b v1.0 is a 12.2B parameter Llama model trained on ChatML and RSLoRA. It is a merge of multiple 12B models and is licensed under apache-2.0. The model is available in several GGUF quantizations.

Good at

  • Reasoning tasks
  • ChatML prompting
  • Large context handling

Limits

  • No vision support
  • Requires at least 8k context for estimates
  • License is apache-2.0

Best for

  • Chatbot development
  • Research prototyping
  • Educational demos

Who it’s for. Developers and researchers who need a reasoning‑capable Llama model with a permissive license. The model is suitable for those who can manage the memory requirements of the chosen quantization.

Quants & memory

QuantFileEst. memoryvs VRAMResultDownload
Q2_K4.8 GB 6.6 GB
Fits Get file →
Q3_K_S5.5 GB 7.4 GB
Tight Get file →
Q3_K_M6.1 GB 7.9 GB
Tight Get file →
Q3_K_L6.6 GB 8.4 GB
Too big Get file →
Q4_07.1 GB 8.9 GB
Too big Get file →
Q4_K_S7.1 GB 9.0 GB
Too big Get file →
Q4_K_M7.5 GB 9.3 GB
Too big Get file →
Q4_17.8 GB 9.6 GB
Too big Get file →
Q5_08.5 GB 10.4 GB
Too big Get file →
Q5_K_S8.5 GB 10.4 GB
Too big Get file →
Q5_K_M8.7 GB 10.6 GB
Too big Get file →
Q5_19.2 GB 11.1 GB
Too big Get file →
Q6_K10.1 GB 11.9 GB
Too big Get file →
Q8_013.0 GB 14.9 GB
Too big Get file →

For 8 GB VRAM, use Q4_K (file size 7.07 GB, estimated 8.9 GB memory). For 12 GB VRAM, Q5_K_M (file size 8.73 GB, estimated 10.6 GB memory) or Q6_K (file size 10.06 GB, estimated 11.9 GB memory). For 16 GB VRAM, Q8_0 (file size 13.02 GB, estimated 14.9 GB memory) offers the highest quality.

What GPU do you need for MN Halide 12b v1.0?

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: the best quant that fits is Q3_K_M (6.1 GB file, about 7.9 GB in use).
  • 12 GB VRAM: the best quant that fits is Q6_K (10.1 GB file, about 11.9 GB in use).
  • 16 GB VRAM: the best quant that fits is Q8_0 (13.0 GB file, about 14.9 GB in use).

Which MN Halide 12b v1.0 quant should you choose?

Q4_K_M is the usual starting point for MN Halide 12b v1.0: it balances file size and quality well. Here it is a 7.5 GB file.

The Q8_0 file is 13.0 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 4.8 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
Parameters12.2B
Layers40
Trained context1,024,000 tokens
KV cache (fp16)160 KB per token

Source

Quantized byQuantFactory
FamilyAzure
Licenseapache-2.0
Downloads (HF)821
Last updated2024-09-24

Frequently asked questions about MN Halide 12b v1.0

How much VRAM does MN Halide 12b v1.0 need?

About 9.3 GB at Q4_K_M with an 8K context window. That is the 7.5 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 MN Halide 12b v1.0 on an 8 GB GPU?

Yes, with Q3_K_M: the 6.1 GB file needs about 7.9 GB at 8K context. Larger quants in this repository do not fit in 8 GB.

What is the file size of MN Halide 12b v1.0 GGUF?

It depends on the quant: Q2_K is 4.8 GB, Q3_K_S is 5.5 GB, Q3_K_M is 6.1 GB, Q3_K_L is 6.6 GB, Q4_0 is 7.1 GB, Q4_K_S is 7.1 GB and 8 more.

What context length does MN Halide 12b v1.0 support?

The GGUF header lists a trained context length of 1,024,000 tokens (1000K). The KV cache grows with the context you use: about 1.3 GB at 8K and 5.4 GB at 32K.

Where can I download MN Halide 12b v1.0 GGUF?

The files are on Hugging Face in the QuantFactory/MN-Halide-12b-v1.0-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 MN Halide 12b v1.0 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 MN Halide 12b v1.0 on 8 GB VRAM?

Yes, using the Q4_K quantization which requires about 8.9 GB of VRAM at 8k context.

Does this model support vision tasks?

No, the model has no vision capability.

What license applies to this model?

The model is released under the apache-2.0 license.

Other GGUFs of Azure_Dusk-v0.2

Similar models

Browse all →

More from QuantFactory

All by QuantFactory →

More Azure models

All Azure →

Other medium models

All medium models →