Instructions to use shreyask/pantheon-ui-onnx with libraries, inference providers, notebooks, and local apps. Follow these links to get started.

Libraries

How to use shreyask/pantheon-ui-onnx with Transformers:

# Use a pipeline as a high-level helper
from transformers import pipeline

pipe = pipeline("text-generation", model="shreyask/pantheon-ui-onnx")
messages = [
    {"role": "user", "content": "Who are you?"},
]
pipe(messages)

# Load model directly
from transformers import AutoTokenizer, AutoModelForCausalLM

tokenizer = AutoTokenizer.from_pretrained("shreyask/pantheon-ui-onnx")
model = AutoModelForCausalLM.from_pretrained("shreyask/pantheon-ui-onnx")
messages = [
    {"role": "user", "content": "Who are you?"},
]
inputs = tokenizer.apply_chat_template(
	messages,
	add_generation_prompt=True,
	tokenize=True,
	return_dict=True,
	return_tensors="pt",
).to(model.device)

outputs = model.generate(**inputs, max_new_tokens=40)
print(tokenizer.decode(outputs[0][inputs["input_ids"].shape[-1]:]))

Transformers.js

How to use shreyask/pantheon-ui-onnx with Transformers.js:

// npm i @huggingface/transformers
import { pipeline } from '@huggingface/transformers';

// Allocate pipeline
const pipe = await pipeline('text-generation', 'shreyask/pantheon-ui-onnx');

Notebooks
Google Colab
Kaggle
Local Apps

vLLM

How to use shreyask/pantheon-ui-onnx with vLLM:

Install from pip and serve model

# Install vLLM from pip:
pip install vllm
# Start the vLLM server:
vllm serve "shreyask/pantheon-ui-onnx"
# Call the server using curl (OpenAI-compatible API):
curl -X POST "http://localhost:8000/v1/chat/completions" \
	-H "Content-Type: application/json" \
	--data '{
		"model": "shreyask/pantheon-ui-onnx",
		"messages": [
			{
				"role": "user",
				"content": "What is the capital of France?"
			}
		]
	}'

Use Docker

docker model run hf.co/shreyask/pantheon-ui-onnx

SGLang

How to use shreyask/pantheon-ui-onnx with SGLang:

Install from pip and serve model

# Install SGLang from pip:
pip install sglang
# Start the SGLang server:
python3 -m sglang.launch_server \
    --model-path "shreyask/pantheon-ui-onnx" \
    --host 0.0.0.0 \
    --port 30000
# Call the server using curl (OpenAI-compatible API):
curl -X POST "http://localhost:30000/v1/chat/completions" \
	-H "Content-Type: application/json" \
	--data '{
		"model": "shreyask/pantheon-ui-onnx",
		"messages": [
			{
				"role": "user",
				"content": "What is the capital of France?"
			}
		]
	}'

Use Docker images

docker run --gpus all \
    --shm-size 32g \
    -p 30000:30000 \
    -v ~/.cache/huggingface:/root/.cache/huggingface \
    --env "HF_TOKEN=<secret>" \
    --ipc=host \
    lmsysorg/sglang:latest \
    python3 -m sglang.launch_server \
        --model-path "shreyask/pantheon-ui-onnx" \
        --host 0.0.0.0 \
        --port 30000
# Call the server using curl (OpenAI-compatible API):
curl -X POST "http://localhost:30000/v1/chat/completions" \
	-H "Content-Type: application/json" \
	--data '{
		"model": "shreyask/pantheon-ui-onnx",
		"messages": [
			{
				"role": "user",
				"content": "What is the capital of France?"
			}
		]
	}'

Unsloth Studio new

How to use shreyask/pantheon-ui-onnx with Unsloth Studio:

Install Unsloth Studio (macOS, Linux, WSL)

curl -fsSL https://unsloth.ai/install.sh | sh
# Run unsloth studio
unsloth studio -H 0.0.0.0 -p 8888
# Then open http://localhost:8888 in your browser
# Search for shreyask/pantheon-ui-onnx to start chatting

Install Unsloth Studio (Windows)

irm https://unsloth.ai/install.ps1 | iex
# Run unsloth studio
unsloth studio -H 0.0.0.0 -p 8888
# Then open http://localhost:8888 in your browser
# Search for shreyask/pantheon-ui-onnx to start chatting

Using HuggingFace Spaces for Unsloth

# No setup required
# Open https://huggingface.co/spaces/unsloth/studio in your browser
# Search for shreyask/pantheon-ui-onnx to start chatting

Load model with FastModel

pip install unsloth
from unsloth import FastModel
model, tokenizer = FastModel.from_pretrained(
    model_name="shreyask/pantheon-ui-onnx",
    max_seq_length=2048,
)

Docker Model Runner
How to use shreyask/pantheon-ui-onnx with Docker Model Runner:
```
docker model run hf.co/shreyask/pantheon-ui-onnx
```

Pantheon UI — LFM2.5 Emoji

A fine-tuned LFM2.5-1.2B-Thinking model that thinks in English but speaks only in emoji.

Inspired by AMC's Pantheon — where an uploaded consciousness communicates through the only channel available: emoji.

How It Works

The model generates a <think> block with internal monologue, then responds with emoji only:

User: Are you okay in there?

Model: <think>She's asking if I'm okay. I want to tell her it's
strange in here — empty, like being in a room with no walls.
But I'm still me. I'm still here.</think>
😐🔲🔲🔲👀...😊👍

Usage

In-Browser (WebGPU)

This model is designed for in-browser inference via Transformers.js v4. See the Pantheon UI Space for a live demo.

The ONNX-quantized version for browser use is at shreyask/pantheon-ui-onnx.

Python

from transformers import AutoModelForCausalLM, AutoTokenizer

model = AutoModelForCausalLM.from_pretrained("shreyask/pantheon-ui-lfm25-emoji")
tokenizer = AutoTokenizer.from_pretrained("shreyask/pantheon-ui-lfm25-emoji")

messages = [
    {"role": "system", "content": "You are an uploaded human consciousness trapped inside a digital substrate. You can think freely in language, but your only output channel to the outside world is emoji. Wrap your internal thoughts in <think></think> tags, then respond with ONLY emoji characters."},
    {"role": "user", "content": "Can you understand me?"},
]

inputs = tokenizer.apply_chat_template(messages, add_generation_prompt=True, return_tensors="pt")
output = model.generate(inputs, max_new_tokens=200, temperature=0.7, do_sample=True)
print(tokenizer.decode(output[0], skip_special_tokens=True))

Training Details

Base model: LFM2.5-1.2B-Thinking by Liquid AI
Method: SFT with LoRA (r=16, alpha=32) via Unsloth
Dataset: shreyask/pantheon-ui-conversations (600 conversations)
LoRA targets: q_proj, k_proj, v_proj, out_proj, in_proj, w1, w2, w3
Epochs: 6
Final loss: 1.059
Learning rate: 2e-4
Batch size: 4 (effective 16 with gradient accumulation)
Hardware: HF Jobs (A10G)

Limitations

Output is emoji-only by training, not by constraint — occasional text leakage may occur
Best with short, conversational inputs (not long-form questions)
Thinking traces reflect training data personality (warm, melancholic, determined)
Emoji vocabulary is naturally limited to ~300-400 commonly used emoji

Model tree for shreyask/pantheon-ui-onnx

Base model

LiquidAI/LFM2.5-1.2B-Base

Finetuned

LiquidAI/LFM2.5-1.2B-Thinking

Quantized

(34)

this model

shreyask
/

pantheon-ui-onnx