What are the hardware requirements for Tulu 30B Fp16?

Tulu 30B Fp16 requires approximately 65.2 GB of VRAM and supports a context window of 2K tokens. Quantized variants may run on less VRAM; see the Quantized Models section on this page.

Who developed Tulu 30B Fp16 and how large is it?

Tulu 30B Fp16 is developed by TheBloke, a model with 30b parameters. The model is published as open weights on Hugging Face and indexed on LLM Explorer with full benchmark history.

How does Tulu 30B Fp16 perform on standard benchmarks?

Tulu 30B Fp16 has the following published scores: MMLU 56.1. Compare against reference models on this page or on the LLM Explorer leaderboards.

Tulu 30B Fp16 by TheBloke — VRAM 65.2GB, 2K context

Name: Tulu 30B Fp16
Author: TheBloke

Tulu 30B Fp16 is an open-source language model by TheBloke. Features: 30b LLM, VRAM: 65.2GB, Context: 2K, License: other, Quantized, LLM Explorer Score: 0.1, Arc: 60, HellaSwag: 83.4, MMLU: 56.1, GSM8K: 19.7.

Arxiv:2301.13688 Arxiv:2302.13971 Arxiv:2304.03277 Arxiv:2304.07327 Arxiv:2306.04751 Dataset:databricks/databricks-... Dataset:openassistant/oasst1 Dataset:sahil2801/codealpaca-2... En Fp16 Llama Pytorch Quantized Region:us Sharded

Model Card on HF 🤗: https://huggingface.co/TheBloke/tulu-30B-fp16

Tulu 30B Fp16 Parameters and Internals

Model Type

Text generation, Instruction tuning

Use Cases

Areas:

Research, Instruction tuning

Applications:

AI-driven text generation, Instruction-based AI applications

Primary Use Cases:

Instruction tuning applications

Considerations:

Use input formatting as prescribed for optimal results.

Supported Languages

en (High proficiency)

Training Details

Data Sources:

databricks/databricks-dolly-15k, OpenAssistant/oasst1, sahil2801/CodeAlpaca-20k, FLAN V2, CoT, Dolly, Open Assistant 1, GPT4-Alpaca, Code-Alpaca, ShareGPT

Methodology:

Finetuned on a mixture of instruction datasets

Model Architecture:

LLaMa 30B architecture

Input Output

Input Format:

<|user|> Your message here! <|assistant|>

Accepted Modalities:

text

Performance Tips:

Ensure newlines are included as prescribed after `<|assistant|>` for proper response.

LLM Name	Tulu 30B Fp16
Repository 🤗	https://huggingface.co/TheBloke/tulu-30B-fp16
Model Size	30b
Required VRAM	65.2 GB
Updated	2026-06-30
Maintainer	TheBloke
Model Type	llama
Model Files	9.8 GB: 1-of-7 10.0 GB: 2-of-7 9.9 GB: 3-of-7 9.9 GB: 4-of-7 9.9 GB: 5-of-7 10.0 GB: 6-of-7 5.7 GB: 7-of-7
Supported Languages	en
Quantization Type	fp16
Model Architecture	LlamaForCausalLM
License	other
Context Length	2048
Model Max Length	2048
Transformers Version	4.30.0.dev0
Tokenizer Class	LlamaTokenizer
Vocabulary Size	32001
Torch Data Type	float16

Rank the Tulu 30B Fp16 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

Best Alternatives to Tulu 30B Fp16

Best Alternatives	Context / RAM	Downloads	Likes
Tenebra 30B Alpha01 FP16	16K / 65 GB	1833	9
Tenebra 30B Alpha01 4BIT	16K / 19.4 GB	73	1
...nebra 30B Alpha01 EXL2 2 80bpw	16K / 11.9 GB	4	1
...nebra 30B Alpha01 EXL2 2 50bpw	16K / 10.7 GB	6	0
Tenebra 30B Alpha01 EXL2 3bpw	16K / 12.7 GB	6	0
Tenebra 30B Alpha01 3BIT	16K / 12.9 GB	5	0
Platypus 30B SuperHOT 8K Fp16	8K / 65.2 GB	176	2
... 30B Supercot SuperHOT 8K Fp16	8K / 65.2 GB	141	4
GPlatty 30B SuperHOT 8K Fp16	8K / 65.2 GB	128	1
Llama 1 30B E8P 2Bit	2K / 8.9 GB	4	1

Note: green Score (e.g. "73.2") means that the model is better than TheBloke/tulu-30B-fp16.