LLM EXPLORER 59,420 MODELS INDEXED

Phi 3 Orpo V8.16 by cstr

By cstr · 7 downloads
HF Repo ↗ Copy URL

Phi 3 Orpo V8.16 is an open-source language model by cstr. Features: LLM, VRAM: 7.6GB, Context: 4K, License: mit, Instruction-Based, LLM Explorer Score: 0.12.

Base model:finetune:vonjack/ph... Base model:vonjack/phi-3-mini-...   Conversational   De   En   Endpoints compatible   Instruct   Llama   Orpo   Pytorch   Region:us   Sharded   Trl   Unsloth
Model Card on HF 🤗: https://huggingface.co/cstr/phi-3-orpo-v8_16 

Phi 3 Orpo V8.16 Parameters and Internals

Model Type 
text-generation-inference, transformers
Additional Notes 
Experiment with Unsloth and Huggingface's TRL library
Supported Languages 
en (unknown), de (unknown)
Training Details 
Data Sources:
German dataset
Methodology:
Only 150 orpo steps
Training Time:
2x faster
Model Architecture:
LLaMA
LLM NamePhi 3 Orpo V8.16
Repository 🤗https://huggingface.co/cstr/phi-3-orpo-v8_16 
Base Model(s)  ...i 3 Mini 4K Instruct LLaMAfied   vonjack/Phi-3-mini-4k-instruct-LLaMAfied
Required VRAM7.6 GB
Updated2026-07-15
Maintainercstr
Model Typellama
Instruction-BasedYes
Model Files  5.0 GB: 1-of-2   2.6 GB: 2-of-2
Supported Languagesen de
Model ArchitectureLlamaForCausalLM
Licensemit
Context Length4096
Model Max Length4096
Transformers Version4.40.0
Tokenizer ClassLlamaTokenizer
Padding Token<|endoftext|>
Vocabulary Size32064
Torch Data Typefloat16

Best Alternatives to Phi 3 Orpo V8.16

Best Alternatives
Context / RAM
Downloads
Likes
Reverse Instruct32K / 27 GB83
Etri Ones Solar4K / 42.9 GB500
Tinyllama Python4K / 2.2 GB132
Small Instruct4K / 2.9 GB101
Taiwan LLaMa V1.04K / 26 GB3578
Taiwan LLaMa V0.04K / 26 GB141
Taiwan LLaMa V0.94K / 26 GB130
Llama2 Ja Small Instruct2K / 1.7 GB51
Kolong Llama V0.12K / 13.7 GB70
...ama Miniguanaco Instruct 121v22K / 2.2 GB110
Note: green Score (e.g. "73.2") means that the model is better than cstr/phi-3-orpo-v8_16.