What are the hardware requirements for Dolphin 2.9.1 Phi 3 Kensho 4.5B AWQ?

Dolphin 2.9.1 Phi 3 Kensho 4.5B AWQ requires approximately 3.7 GB of VRAM and supports a context window of 4K tokens. Quantized variants may run on less VRAM; see the Quantized Models section on this page.

Who developed Dolphin 2.9.1 Phi 3 Kensho 4.5B AWQ and how large is it?

Dolphin 2.9.1 Phi 3 Kensho 4.5B AWQ is developed by vaclavkosar, a model with 4.5b 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 2.9.1 Phi 3 Kensho 4.5B AWQ?

Dolphin 2.9.1 Phi 3 Kensho 4.5B AWQ is hosted on Hugging Face and linked from this page. LLM Explorer also lists quantized variants and similar alternatives if available.

Dolphin 2.9.1 Phi 3 Kensho 4.5B AWQ by vaclavkosar — VRAM 3.7GB, 4K context

Name: Dolphin 2.9.1 Phi 3 Kensho 4.5B AWQ
Author: vaclavkosar

Dolphin 2.9.1 Phi 3 Kensho 4.5B AWQ is an open-source language model by vaclavkosar. Features: 4.5b LLM, VRAM: 3.7GB, Context: 4K, License: mit, Quantized, Instruction-Based, LLM Explorer Score: 0.12.

4-bit Awq Base model:quantized:unsloth/p... Base model:unsloth/phi-3-mini-... 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 En Endpoints compatible Instruct Mistral Quantized Region:us Safetensors

Model Card on HF 🤗: https://huggingface.co/vaclavkosar/Dolphin-2.9.1-Phi-3-Kensho-4.5B-AWQ

Dolphin 2.9.1 Phi 3 Kensho 4.5B AWQ Parameters and Internals

Model Type

Instruction, Conversational, Coding

Additional Notes

The dataset was filtered to remove alignment and bias, making the model more compliant. An additional alignment layer is advised before deploying as a service.

Supported Languages

en (English)

Training Details

Data Sources:

cognitivecomputations/Dolphin-2.9, teknium/OpenHermes-2.5, m-a-p/CodeFeedback-Filtered-Instruction, cognitivecomputations/dolphin-coder, cognitivecomputations/samantha-data, microsoft/orca-math-word-problems-200k, Locutusque/function-calling-chatml, internlm/Agent-FLAN

Methodology:

This model utilizes PEFT layer replication at inference time to duplicate layers and increase parameter count. The qLoRA fine-tuning was with 4k sequence length.

Context Length:

4000

Training Time:

2.5 days

Hardware Used:

8xL40S node provided by Crusoe Cloud

Input Output

Input Format:

ChatML prompt template format

LLM Name	Dolphin 2.9.1 Phi 3 Kensho 4.5B AWQ
Repository 🤗	https://huggingface.co/vaclavkosar/Dolphin-2.9.1-Phi-3-Kensho-4.5B-AWQ
Base Model(s)	Phi 3 Mini 4K Instruct unsloth/Phi-3-mini-4k-instruct
Model Size	4.5b
Required VRAM	3.7 GB
Updated	2026-06-03
Maintainer	vaclavkosar
Model Type	mistral
Instruction-Based	Yes
Model Files	3.7 GB
Supported Languages	en
AWQ Quantization	Yes
Quantization Type	awq
Model Architecture	MistralForCausalLM
License	mit
Context Length	4096
Model Max Length	4096
Transformers Version	4.41.0
Tokenizer Class	LlamaTokenizer
Padding Token	<\|endoftext\|>
Vocabulary Size	32064
Torch Data Type	float16

Rank the Dolphin 2.9.1 Phi 3 Kensho 4.5B AWQ 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 2.9.1 Phi 3 Kensho 4.5B AWQ

Best Alternatives	Context / RAM	Downloads	Likes
...olphin 2.9.1 Phi 3 Kensho 4.5B	4K / 7.6 GB	16	32
...i 3 Kensho 4.5B Abliterated V3	4K / 13.1 GB	63	10
...i 3 Kensho 4.5B Abliterated V3	4K / 13.1 GB	12	10
...olphin 2.9.1 Phi 3 Kensho 4.5B	4K / 7.6 GB	45	31

Note: green Score (e.g. "73.2") means that the model is better than vaclavkosar/Dolphin-2.9.1-Phi-3-Kensho-4.5B-AWQ.