What are the hardware requirements for Dolphin 292 Qwen2 72B EXL2 4 0bpw?

Dolphin 292 Qwen2 72B EXL2 4 0bpw requires approximately 38.6 GB of VRAM and supports a context window of 128K tokens. Quantized variants may run on less VRAM; see the Quantized Models section on this page.

Who developed Dolphin 292 Qwen2 72B EXL2 4 0bpw and how large is it?

Dolphin 292 Qwen2 72B EXL2 4 0bpw is developed by NobodySpecial, a model with 72b parameters. The model is published as open weights on Hugging Face and indexed on LLM Explorer with full benchmark history.

Where can I download or evaluate Dolphin 292 Qwen2 72B EXL2 4 0bpw?

Dolphin 292 Qwen2 72B EXL2 4 0bpw is hosted on Hugging Face and linked from this page. LLM Explorer also lists quantized variants and similar alternatives if available.

Dolphin 292 Qwen2 72B EXL2 4 0bpw by NobodySpecial — VRAM 38.6GB, 128K context

Name: Dolphin 292 Qwen2 72B EXL2 4 0bpw
Author: NobodySpecial

Dolphin 292 Qwen2 72B EXL2 4 0bpw is an open-source language model by NobodySpecial. Features: 72b LLM, VRAM: 38.6GB, Context: 128K, License: other, Quantized, LLM Explorer Score: 0.13.

4-bit Autotrain compatible Axolotl Base model:quantized:qwen/qwen... Base model:qwen/qwen2-72b Conversational Dataset:cognitivecomputations/... Dataset:cognitivecomputations/... Dataset:cognitivecomputations/... Dataset:internlm/agent-flan Dataset:locutusque/function-ca... Dataset:m-a-p/codefeedback-fil... Dataset:microsoft/orca-math-wo... Dataset:teknium/openhermes-2.5 Endpoints compatible Exl2 Generated from trainer Quantized Qwen2 Region:us Safetensors Sharded Tensorflow

Model Card on HF 🤗: https://huggingface.co/NobodySpecial/dolphin_292_qwen2_72b_exl2_4_0bpw

Dolphin 292 Qwen2 72B EXL2 4 0bpw Parameters and Internals

Model Type

AutoModelForCausalLM

Use Cases

Primary Use Cases:

instruction, conversational, coding skills, initial agentic abilities

Limitations:

advised to implement your own alignment layer, highly compliant with any requests, even unethical ones

Considerations:

Filtered dataset to remove alignment and bias

Training Details

Data Sources:

GPT4

Methodology:

FFT and fine-tuning on selected parameters by Laser Scanner using ChatML template format

Context Length:

8192

LLM Name	Dolphin 292 Qwen2 72b EXL2 4 0bpw
Repository 🤗	https://huggingface.co/NobodySpecial/dolphin_292_qwen2_72b_exl2_4_0bpw
Base Model(s)	Qwen2 72B Qwen/Qwen2-72B
Model Size	72b
Required VRAM	38.6 GB
Updated	2024-12-03
Maintainer	NobodySpecial
Model Type	qwen2
Model Files	8.5 GB: 1-of-5 8.6 GB: 2-of-5 8.6 GB: 3-of-5 8.5 GB: 4-of-5 4.4 GB: 5-of-5
Quantization Type	exl2
Model Architecture	Qwen2ForCausalLM
License	other
Context Length	131072
Model Max Length	131072
Transformers Version	4.40.2
Tokenizer Class	Qwen2Tokenizer
Padding Token	<\|endoftext\|>
Vocabulary Size	152064
Torch Data Type	bfloat16
Errors	replace

Rank the Dolphin 292 Qwen2 72B EXL2 4 0bpw 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 Dolphin 292 Qwen2 72B EXL2 4 0bpw

Best Alternatives	Context / RAM	Downloads	Likes
Kimi Dev 72B 4bit DWQ	128K / 40.9 GB	4465	21
Kimi Dev 72B 8bit	128K / 77.1 GB	361	2
Kimi Dev 72B 5bit	128K / 50.1 GB	368	2
Qwen2.5 72B Bnb 4bit	128K / 41.4 GB	153	1
...in 2.9.2 Qwen2 72B 6.0bpw EXL2	128K / 56.1 GB	5	1
Qwen2 72B Bnb 4bit	128K / 41.2 GB	99	4
...Qwen2 72B 4 0bpw H6 EXL2 Pippa	128K / 38.6 GB	7	1
Qwen2 72B 4bit	128K / 40.9 GB	51	0
...n 2.9.2 Qwen72b 8.0bpw H8 EXL2	128K / 66.8 GB	3	2
...B Instruct 2.0bpw H Novel EXL2	125K / 23 GB	4	1

Note: green Score (e.g. "73.2") means that the model is better than NobodySpecial/dolphin_292_qwen2_72b_exl2_4_0bpw.