What are the hardware requirements for CodeLlama 13B Hf?

CodeLlama 13B Hf requires approximately 26 GB of VRAM and supports a context window of 16K tokens. Quantized variants may run on less VRAM; see the Quantized Models section on this page.

Who developed CodeLlama 13B Hf and how large is it?

CodeLlama 13B Hf is developed by codellama, a model with 13b parameters. The model is published as open weights on Hugging Face and indexed on LLM Explorer with full benchmark history.

How does CodeLlama 13B Hf perform on standard benchmarks?

CodeLlama 13B Hf has the following published scores: MMLU 32.81, HumanEval (Python) 35.07. Compare against reference models on this page or on the LLM Explorer leaderboards.

CodeLlama 13B Hf by codellama — VRAM 26GB, 16K context

Name: CodeLlama 13B Hf
Author: codellama

CodeLlama 13B Hf is an open-source language model by codellama. Features: 13b LLM, VRAM: 26GB, Context: 16K, License: llama2, Code Generating, LLM Explorer Score: 0.14, Arc: 40.9, HellaSwag: 63.4, MMLU: 32.8, GSM8K: 12.1, HumanEval: 35.1.

Arxiv:2308.12950 Code Codegen Deploy:azure Deploy:sagemaker Endpoints compatible Llama Llama2 Pytorch Region:us Safetensors Sharded Tensorflow

Model Card on HF 🤗: https://huggingface.co/codellama/CodeLlama-13b-hf

CodeLlama 13B Hf Parameters and Internals

Model Type

text-generation

Use Cases

Areas:

commercial, research

Applications:

code synthesis, understanding tasks, handling Python programming

Primary Use Cases:

code assistant and generation applications

Limitations:

Use in languages other than English, Using violating laws or regulations

Considerations:

Follow Acceptable Use Policy and Licensing Agreement

Additional Notes

This is a non-official Code Llama repo. Official repository available at Meta's Llama organization on Hugging Face.

Supported Languages

English (high proficiency)

Training Details

Data Sources:

same data as Llama 2

Data Volume:

Not specified

Methodology:

Pre-trained and fine-tuned generative text models

Training Time:

January 2023 to July 2023

Hardware Used:

Meta’s Research Super Cluster with A100-80GB GPUs

Model Architecture:

auto-regressive language model using an optimized transformer architecture

Safety Evaluation

Methodologies:

community feedback

Risk Categories:

misinformation, bias

Ethical Considerations:

Potential to produce inaccurate or objectionable responses

Responsible Ai Considerations

Fairness:

Testing conducted only in English; may not cover all scenarios

Transparency:

Outputs cannot be predicted in advance

Accountability:

Developers should perform safety testing and application tuning

Mitigation Strategies:

Responsible Use Guide available at Meta's website

Input Output

Input Format:

text only

Accepted Modalities:

text

Output Format:

text only

LLM Name	CodeLlama 13B Hf
Repository 🤗	https://huggingface.co/codellama/CodeLlama-13b-hf
Model Size	13b
Required VRAM	26 GB
Updated	2026-07-25
Maintainer	codellama
Model Type	llama
Model Files	9.9 GB: 1-of-3 9.9 GB: 2-of-3 6.2 GB: 3-of-3 9.9 GB: 1-of-3 9.9 GB: 2-of-3 6.5 GB: 3-of-3
Supported Languages	code
Generates Code	Yes
Model Architecture	LlamaForCausalLM
License	llama2
Context Length	16384
Model Max Length	16384
Transformers Version	4.32.0.dev0
Tokenizer Class	CodeLlamaTokenizer
Beginning of Sentence Token	<s>
End of Sentence Token	</s>
Unk Token	<unk>
Vocabulary Size	32016
Torch Data Type	bfloat16

Rank the CodeLlama 13B Hf capabilities

Have you tried this model? Rate its performance — this feedback helps the ML community find the right model for their needs.

Instruction Following and Task Automation
Factuality and Completeness of Knowledge
Censorship and Alignment
Data Analysis and Insight Generation
Text Generation
Text Summarization and Feature Extraction
Code Generation
Multi-Language Support and Translation

Quantized Models of the CodeLlama 13B Hf

Model	Likes	Downloads	VRAM
CodeLlama 13B GGUF	60	1262	5 GB
CodeLlama 13B AWQ	4	115	7 GB
CodeLlama 13B GPTQ	12	62	7 GB
CodeLlama 13B GGML	33	11	5 GB

Best Alternatives to CodeLlama 13B Hf

Best Alternatives	Context / RAM	Downloads	Likes
NexusRaven V2 13B	16K / 26 GB	105	472
CodeLlama 13B Instruct Hf	16K / 26 GB	2469	160
CodeLlama 13B MORepair	16K / 26 GB	9	2
Tora Code 13B V1.0	16K / 26 GB	878	15
CodeLlama 13B Python Hf	16K / 26 GB	558	51
...ma 13B Hf Truncated Embeddings	16K / 52.3 GB	5	0
...ma Airoboros Orca Platypus 13B	16K / 26 GB	117	0
Codellama 13B Oasst Sft V10	16K / 25.9 GB	184	70
CodeLlama 13B Hf	16K / 26 GB	224	0
CodeLlama 13B Instruct Hf	16K / 26 GB	369	29

Note: green Score (e.g. "73.2") means that the model is better than codellama/CodeLlama-13b-hf.