LL

CodeLlama 13b Hf

LlamaGGUF by mradermacherUpdated 2024-09-11
Download on Hugging Face
Parameters13B
Context16K
Layers40
Architecturellama
Smallest quant2.9 GB
Quick answer

CodeLlama 13b Hf 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 21 quants, from 2.9 GB (IQ1_S) to 10.7 GB (Q6_K).

CodeLlama 13B Hf – a 13‑billion‑parameter code generation model

CodeLlama 13B Hf is a 13‑billion‑parameter Llama‑based model fine‑tuned for code synthesis and understanding. It supports code completion and infilling but not instruction or chat. The model is available in multiple GGUF quantizations.

Good at

  • Code completion
  • Code infilling
  • General code synthesis

Limits

  • No instruction or chat support
  • Requires 8‑k context for memory estimates
  • License is llama2

Best for

  • Writing boilerplate code
  • Debugging assistance
  • Generating code snippets

Who it’s for. Developers and researchers needing a large language model focused on code tasks. The base model can be adapted for various programming languages, but it is not specialized for Python or instruction following.

Quants & memory

QuantFileEst. memoryvs VRAMResultDownload
IQ1_S2.9 GB 10.1 GB
Too big Get file →
IQ1_M3.1 GB 10.3 GB
Too big Get file →
IQ2_XXS3.5 GB 10.7 GB
Too big Get file →
IQ2_XS3.9 GB 11.1 GB
Too big Get file →
IQ2_S4.2 GB 11.4 GB
Too big Get file →
IQ2_M4.5 GB 11.7 GB
Too big Get file →
Q2_K4.9 GB 12.1 GB
Too big Get file →
IQ3_XXS5.0 GB 12.2 GB
Too big Get file →
IQ3_XS5.4 GB 12.6 GB
Too big Get file →
IQ3_S5.7 GB 12.9 GB
Too big Get file →
Q3_K_S5.7 GB 12.9 GB
Too big Get file →
IQ3_M6.0 GB 13.2 GB
Too big Get file →
Q3_K_M6.3 GB 13.5 GB
Too big Get file →
Q3_K_L6.9 GB 14.1 GB
Too big Get file →
IQ4_XS7.0 GB 14.2 GB
Too big Get file →
Q4_07.4 GB 14.6 GB
Too big Get file →
Q4_K_S7.4 GB 14.6 GB
Too big Get file →
Q4_K_M7.9 GB 15.1 GB
Too big Get file →
Q5_K_S9.0 GB 16.2 GB
Too big Get file →
Q5_K_M9.2 GB 16.4 GB
Too big Get file →
Q6_K10.7 GB 17.9 GB
Too big Get file →

For 8‑GB VRAM use IQ1_S (2.9 GB, 10.1 GB memory) or IQ1_M (3.14 GB, 10.3 GB memory). With 16‑GB VRAM, Q4_K_M (7.87 GB, 15.1 GB memory) or Q5_K_S (8.97 GB, 16.2 GB memory) are suitable. Higher VRAM allows larger quants such as Q6_K (10.68 GB, 17.9 GB memory).

What GPU do you need for CodeLlama 13b Hf?

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: the best quant that fits is IQ2_M (4.5 GB file, about 11.7 GB in use).
  • 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 Q6_K (10.7 GB file, about 17.9 GB in use).

Which CodeLlama 13b Hf quant should you choose?

Q4_K_M is the usual starting point for CodeLlama 13b Hf: it balances file size and quality well. Here it is a 7.9 GB file.

If memory is tight, the smallest file is IQ1_S at 2.9 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
Parameters13B
Layers40
Trained context16,384 tokens
KV cache (fp16)800 KB per token

Source

Quantized bymradermacher
FamilyLlama
Licensellama2
Downloads (HF)3,042
Last updated2024-09-11

Frequently asked questions about CodeLlama 13b Hf

How much VRAM does CodeLlama 13b Hf 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 CodeLlama 13b Hf on an 8 GB GPU?

Not at an 8K context. The smallest file needs about 10.1 GB, so you would need a larger GPU or CPU offloading.

What is the file size of CodeLlama 13b Hf GGUF?

It depends on the quant: IQ1_S is 2.9 GB, IQ1_M is 3.1 GB, IQ2_XXS is 3.5 GB, IQ2_XS is 3.9 GB, IQ2_S is 4.2 GB, IQ2_M is 4.5 GB and 15 more.

What context length does CodeLlama 13b Hf support?

The GGUF header lists a trained context length of 16,384 tokens (16K). The KV cache grows with the context you use: about 6.7 GB at 8K and 13.4 GB at 16K.

Where can I download CodeLlama 13b Hf GGUF?

The files are on Hugging Face in the mradermacher/CodeLlama-13b-hf-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 CodeLlama 13b Hf use?

Hugging Face lists the license as llama2. 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 CodeLlama 13B Hf on 8 GB VRAM?

Yes, using IQ1_S or IQ1_M quantizations which require about 10 GB of VRAM at 8‑k context.

Does this model support chat or instruction following?

No, the base model supports only code completion and infilling, not instruction or chat.

What license applies to this model?

The model is governed by the llama2 license.

Other GGUFs of CodeLlama-13b-hf

Similar models

Browse all →

More from mradermacher

All by mradermacher →

More Llama models

All Llama →

Other medium models

All medium models →