A Unity package to run pretrained transformer models with Unity Sentis
C#
25
156 commits
updated Apr 4, 2026
A Unity package to run pretrained transformer models with Unity Sentis
OpenUPM · Documentation (coming soon) · Feedback/Questions
This is essentially a C# port of Hugging Face’s transformers library.
There are two use cases for this package right now:
In Edit -> Project Settings -> Package Manager, add a new scoped registry:
Name: Doji
URL: https://package.openupm.com
Scope(s): com.doji
In the Package Manager install com.doji.transformers either by name or select it in the list under Package Manager -> My Registries
Tokenizers
LLMs
The intention is to provide a similar API like Hugging Face's transformers library, so usage in Unity will look something like this:
var tokenizer = AutoTokenizer.FromPretrained("julienkay/Phi-3-mini-4k-instruct_no_cache_uint8");
var model = Phi3ForCausalLM.FromPretrained("julienkay/Phi-3-mini-4k-instruct_no_cache_uint8");
var inputs = tokenizer.Encode("<input-prompt>");
var outputs = model.Generate(inputs);
var predictedText = tokenizer.Decode(outputs);
C#
100.0%
A Unity package to run pretrained transformer models with Unity Sentis
C#
25
156 commits
updated Apr 4, 2026
A Unity package to run pretrained transformer models with Unity Sentis
OpenUPM · Documentation (coming soon) · Feedback/Questions
This is essentially a C# port of Hugging Face’s transformers library.
There are two use cases for this package right now:
In Edit -> Project Settings -> Package Manager, add a new scoped registry:
Name: Doji
URL: https://package.openupm.com
Scope(s): com.doji
In the Package Manager install com.doji.transformers either by name or select it in the list under Package Manager -> My Registries
Tokenizers
LLMs
The intention is to provide a similar API like Hugging Face's transformers library, so usage in Unity will look something like this:
var tokenizer = AutoTokenizer.FromPretrained("julienkay/Phi-3-mini-4k-instruct_no_cache_uint8");
var model = Phi3ForCausalLM.FromPretrained("julienkay/Phi-3-mini-4k-instruct_no_cache_uint8");
var inputs = tokenizer.Encode("<input-prompt>");
var outputs = model.Generate(inputs);
var predictedText = tokenizer.Decode(outputs);
C#
100.0%