AL

ALMA 13B Pretrain

AlmaGGUF by TheBlokeUpdated 2023-09-27
Download on Hugging Face
Parameters13B
Context4K
Layers40
Architecturellama
Smallest quant5.4 GB
Quick answer

ALMA 13B Pretrain needs about 15.1 GB of memory at Q4_K_M (a 7.9 GB file) with an 8K context window, so it fits on a 16 GB GPU. This repository has 12 quants, from 5.4 GB (Q2_K) to 13.8 GB (Q8_0).

ALMA 13B Pretrain – a 13B parameter LLM fine‑tuned on monolingual data

ALMA 13B Pretrain is a 13 billion parameter language model based on the LLaMA architecture. It has been fine‑tuned on 12 billion monolingual tokens and is released in GGUF format for efficient inference. The model supports a 4096‑token context window and is licensed under MIT.

Good at

  • Monolingual fine‑tuning
  • Large context handling
  • Low‑memory inference

Limits

  • No translation capability without LoRA
  • Requires 8 k context for memory estimates
  • No vision support

Best for

  • Local inference on modest GPUs
  • Research on monolingual language modeling
  • Fine‑tuning with LoRA adapters

Who it’s for. Researchers and developers who need a lightweight, open‑source 13 billion parameter model for local deployment. The model is ideal for building custom adapters or exploring monolingual fine‑tuning workflows.

Quants & memory

QuantFileEst. memoryvs VRAMResultDownload
Q2_K5.4 GB 9.3 GB
Too big Get file →
Q3_K_S5.7 GB 9.5 GB
Too big Get file →
Q3_K_M6.3 GB 10.2 GB
Too big Get file →
Q3_K_L6.9 GB 10.8 GB
Too big Get file →
Q4_07.4 GB 11.2 GB
Too big Get file →
Q4_K_S7.4 GB 11.3 GB
Too big Get file →
Q4_K_M7.9 GB 11.7 GB
Too big Get file →
Q5_09.0 GB 12.8 GB
Too big Get file →
Q5_K_S9.0 GB 12.8 GB
Too big Get file →
Q5_K_M9.2 GB 13.1 GB
Too big Get file →
Q6_K10.7 GB 14.5 GB
Too big Get file →
Q8_013.8 GB 17.7 GB
Too big Get file →

For 8 GB VRAM, use Q4_K_M (7.87 GB file, ~15.1 GB VRAM at 8 k context). For 12 GB VRAM, Q5_K_M (9.23 GB file, ~16.4 GB VRAM) offers a balance of quality and memory. Higher quants such as Q6_K or Q8_0 require 16 GB+ VRAM.

What GPU do you need for ALMA 13B Pretrain?

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: no quant fits at an 8K context.
  • 16 GB VRAM: the best quant that fits is Q4_K_M (7.9 GB file, about 15.1 GB in use).
  • 24 GB VRAM: the best quant that fits is Q8_0 (13.8 GB file, about 21.0 GB in use).

Which ALMA 13B Pretrain quant should you choose?

Q4_K_M is the usual starting point for ALMA 13B Pretrain: it balances file size and quality well. Here it is a 7.9 GB file.

The Q8_0 file is 13.8 GB, about 1.8 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 5.4 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
Parameters13B
Layers40
Trained context4,096 tokens
KV cache (fp16)800 KB per token

Source

Quantized byTheBloke
FamilyAlma
Licensemit
Downloads (HF)3,124
Last updated2023-09-27

Frequently asked questions about ALMA 13B Pretrain

How much VRAM does ALMA 13B Pretrain need?

About 15.1 GB at Q4_K_M with an 8K context window. That is the 7.9 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 ALMA 13B Pretrain on an 8 GB GPU?

Not at an 8K context. The smallest file needs about 12.6 GB, so you would need a larger GPU or CPU offloading.

What is the file size of ALMA 13B Pretrain GGUF?

It depends on the quant: Q2_K is 5.4 GB, Q3_K_S is 5.7 GB, Q3_K_M is 6.3 GB, Q3_K_L is 6.9 GB, Q4_0 is 7.4 GB, Q4_K_S is 7.4 GB and 6 more.

What context length does ALMA 13B Pretrain support?

The GGUF header lists a trained context length of 4,096 tokens (4K). The KV cache grows with the context you use: about 6.7 GB at 8K and 3.4 GB at 4K.

Where can I download ALMA 13B Pretrain GGUF?

The files are on Hugging Face in the TheBloke/ALMA-13B-Pretrain-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 ALMA 13B Pretrain 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 ALMA 13B Pretrain on 8 GB VRAM?

Use the Q4_K_M quantized file; it requires about 15.1 GB VRAM at 8 k context, so 8 GB VRAM is insufficient.

Does this model support translation?

No. ALMA 13B Pretrain is only the monolingual pre‑train stage; translation requires the LoRA or R variants.

What license applies to this model?

The model is released under the MIT license, allowing free use and modification.

Other GGUFs of ALMA-13B-Pretrain

Similar models

Browse all →

More from TheBloke

All by TheBloke →

More Alma models

All Alma →

Other medium models

All medium models →