What are the hardware requirements for Phi 2 Orange?

Phi 2 Orange requires approximately 5.6 GB of VRAM. Quantized variants may run on less VRAM; see the Quantized Models section on this page.

Who developed Phi 2 Orange and how large is it?

Phi 2 Orange is developed by rhysjones, 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?

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

Phi 2 Orange by rhysjones — VRAM 5.6GB

Name: Phi 2 Orange
Author: rhysjones

Phi 2 Orange is an open-source language model by rhysjones. Features: 2.8b LLM, VRAM: 5.6GB, License: mit, LLM Explorer Score: 0.1.

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... Endpoints compatible Phi-msft Region:us Safetensors Sharded Tensorflow

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

Phi 2 Orange Parameters and Internals

Model Type

text generation

Additional Notes

An updated model is available at [rhysjones/phi-2-orange-v2](https://huggingface.co/rhysjones/phi-2-orange-v2) with higher evaluations.

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 fine-tuning: a collection of broad training data followed by DPO fine-tuning

Input Output

Input Format:

ChatML

Accepted Modalities:

text

Output Format:

Markdown format for Python outputs

LLM Name	Phi 2 Orange
Repository 🤗	https://huggingface.co/rhysjones/phi-2-orange
Model Size	2.8b
Required VRAM	5.6 GB
Updated	2026-05-27
Maintainer	rhysjones
Model Type	phi-msft
Model Files	5.0 GB: 1-of-2 0.6 GB: 2-of-2
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 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 Phi 2 Orange

Model	Likes	Downloads	VRAM
Phi 2 Orange GGUF	20	591	1 GB
Phi 2 Orange GPTQ	4	17	1 GB

Best Alternatives to Phi 2 Orange

Best Alternatives	Context / RAM	Downloads
MFANN3bv0.24	128K / 11.1 GB	6
MFANN3b	128K / 11.1 GB	86
MFANN3bv1.4	128K / 11.1 GB	8
MFANN Phigments Slerp V3.2	128K / 5.6 GB	6
MFANN3bv1.3	128K / 11.1 GB	14
MFANN3bv0.23	128K / 11.1 GB	10
MFANN3bv1.1	128K / 11.1 GB	10
MFANN3bv1.5	128K / 11.1 GB	8
MFANN Liminerity Slerp 4a	128K / 5.6 GB	7
MFANN3b Rebase	128K / 11.1 GB	7

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