What are the hardware requirements for Phi 2 Orange GPTQ?

Phi 2 Orange GPTQ requires approximately 1.8 GB of VRAM. Quantized variants may run on less VRAM; see the Quantized Models section on this page.

Who developed Phi 2 Orange GPTQ and how large is it?

Phi 2 Orange GPTQ is developed by TheBloke, a model with 2.8b 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 Phi 2 Orange GPTQ?

Phi 2 Orange GPTQ is hosted on Hugging Face and linked from this page. LLM Explorer also lists quantized variants and similar alternatives if available.

Phi 2 Orange GPTQ by TheBloke — VRAM 1.8GB

Name: Phi 2 Orange GPTQ
Author: TheBloke

Phi 2 Orange GPTQ is an open-source language model by TheBloke. Features: 2.8b LLM, VRAM: 1.8GB, License: mit, Quantized, LLM Explorer Score: 0.1.

4-bit Base model:quantized:rhysjones... Base model:rhysjones/phi-2-ora... Custom code Dataset:argilla/ultrafeedback-... Dataset:intel/orca dpo pairs Dataset:ldjnr/capybara Dataset:ldjnr/pure-dove Dataset:ldjnr/verified-camel Dataset:meta-math/metamathqa Dataset:migtissera/synthia-v1.... Dataset:open-orca/slimorca-ded... Gptq Phi-msft Quantized Region:us Safetensors

Model Card on HF 🤗: https://huggingface.co/TheBloke/phi-2-orange-GPTQ

Phi 2 Orange GPTQ Parameters and Internals

Model Type

phi-msft

Additional Notes

This model is presented with multiple GPTQ parameter permutations for varying hardware requirements and performance. Quantization is largely supported on GPUs, with some parameters available for both GPUs and CPUs.

Training Details

Data Sources:

Open-Orca/SlimOrca-Dedup, migtissera/Synthia-v1.3, LDJnr/Verified-Camel, LDJnr/Pure-Dove, LDJnr/Capybara, meta-math/MetaMathQA, Intel/orca_dpo_pairs, argilla/ultrafeedback-binarized-preferences-cleaned

Methodology:

Two-step finetune of Phi-2 with additional DPO finetuning

Model Architecture:

Phi-2

Input Output

Input Format:

{prompt}

Output Format:

Unknown

LLM Name	Phi 2 Orange GPTQ
Repository 🤗	https://huggingface.co/TheBloke/phi-2-orange-GPTQ
Model Name	Phi 2 Orange
Model Creator	Rhys Jones
Base Model(s)	Phi 2 Orange rhysjones/phi-2-orange
Model Size	2.8b
Required VRAM	1.8 GB
Updated	2026-05-20
Maintainer	TheBloke
Model Type	phi-msft
Model Files	1.8 GB
GPTQ Quantization	Yes
Quantization Type	gptq
Model Architecture	PhiForCausalLM
License	mit
Model Max Length	2048
Transformers Version	4.37.0.dev0
Tokenizer Class	CodeGenTokenizer
Padding Token	<\|endoftext\|>
Vocabulary Size	51200
Torch Data Type	bfloat16
Activation Function	gelu_new

Rank the Phi 2 Orange GPTQ 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 Phi 2 Orange GPTQ

Best Alternatives	Context / RAM	Downloads	Likes
Phi 2 Quantize Gptq	2K / 1.8 GB	6	2
... 2 Electrical Engineering GPTQ	2K / 1.8 GB	9	7
Phi2 OSST GPTQ	2K / 1.8 GB	9	1
Phi 2 DPO GPTQ	0K / 1.8 GB	37	5
Phi 2 GPTQ	0K / 1.8 GB	65	30
Dolphin 2 6 Phi 2 GPTQ	0K / 1.8 GB	5	2
LMCocktail Phi 2 V1 GPTQ	0K / 1.8 GB	6	1
Bnb DPO 8bit	2K / 3 GB	7	0
Bnb DPO 8bit	2K / 3 GB	7	0
Phi 2 4bit 64rank	2K / 5.6 GB	13	0

Note: green Score (e.g. "73.2") means that the model is better than TheBloke/phi-2-orange-GPTQ.