Instructions to use MicheRomChis/micro-terse with libraries, inference providers, notebooks, and local apps. Follow these links to get started.

Notebooks
Google Colab
Kaggle
Local Apps Settings

How to use MicheRomChis/micro-terse with llama.cpp:

Install (macOS, Linux)

curl -LsSf https://llama.app/install.sh | sh
# Start a local OpenAI-compatible server with a web UI:
llama serve -hf MicheRomChis/micro-terse:TQ2_0
# Run inference directly in the terminal:
llama cli -hf MicheRomChis/micro-terse:TQ2_0

Install from WinGet (Windows)

winget install llama.cpp
# Start a local OpenAI-compatible server with a web UI:
llama serve -hf MicheRomChis/micro-terse:TQ2_0
# Run inference directly in the terminal:
llama cli -hf MicheRomChis/micro-terse:TQ2_0

Use pre-built binary

# Download pre-built binary from:
# https://github.com/ggerganov/llama.cpp/releases
# Start a local OpenAI-compatible server with a web UI:
./llama-server -hf MicheRomChis/micro-terse:TQ2_0
# Run inference directly in the terminal:
./llama-cli -hf MicheRomChis/micro-terse:TQ2_0

Build from source code

git clone https://github.com/ggerganov/llama.cpp.git
cd llama.cpp
cmake -B build
cmake --build build -j --target llama-server llama-cli
# Start a local OpenAI-compatible server with a web UI:
./build/bin/llama-server -hf MicheRomChis/micro-terse:TQ2_0
# Run inference directly in the terminal:
./build/bin/llama-cli -hf MicheRomChis/micro-terse:TQ2_0

Use Docker

docker model run hf.co/MicheRomChis/micro-terse:TQ2_0

LM Studio
Jan

vLLM

How to use MicheRomChis/micro-terse with vLLM:

Install from pip and serve model

# Install vLLM from pip:
pip install vllm
# Start the vLLM server:
vllm serve "MicheRomChis/micro-terse"
# Call the server using curl (OpenAI-compatible API):
curl -X POST "http://localhost:8000/v1/completions" \
	-H "Content-Type: application/json" \
	--data '{
		"model": "MicheRomChis/micro-terse",
		"prompt": "Once upon a time,",
		"max_tokens": 512,
		"temperature": 0.5
	}'

Use Docker

docker model run hf.co/MicheRomChis/micro-terse:TQ2_0

Ollama
How to use MicheRomChis/micro-terse with Ollama:
```
ollama run hf.co/MicheRomChis/micro-terse:TQ2_0
```

Unsloth Studio

How to use MicheRomChis/micro-terse 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 MicheRomChis/micro-terse 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 MicheRomChis/micro-terse to start chatting

Using HuggingFace Spaces for Unsloth

# No setup required
# Open https://huggingface.co/spaces/unsloth/studio in your browser
# Search for MicheRomChis/micro-terse to start chatting

Docker Model Runner
How to use MicheRomChis/micro-terse with Docker Model Runner:
```
docker model run hf.co/MicheRomChis/micro-terse:TQ2_0
```

Lemonade

How to use MicheRomChis/micro-terse with Lemonade:

Pull the model

# Download Lemonade from https://lemonade-server.ai/
lemonade pull MicheRomChis/micro-terse:TQ2_0

Run and chat with the model

lemonade run user.micro-terse-TQ2_0

List all available models

lemonade list

Atomic Chat

MicheRomChis commited on Jun 26

Commit

87d72e0

verified ·

1 Parent(s): 186292c

Accuracy pass: Q6_K embedding note, fixed tau, filenames

Browse files

Files changed (1) hide show

README.md +2 -2

README.md CHANGED Viewed

@@ -32,7 +32,7 @@ tags:
 **Micro-Terse** is a 423M-parameter (≈320M active) **ternary-weight** language model trained from
 scratch for ≈**$150**, deployable as a **182 MB CPU-only GGUF**. Its weights are constrained to
-`{−1, 0, +1}` (≈1.58 bits), so `TQ2_0` packing is lossless.
 It is a research proof-of-concept, **not** a production assistant. At an 8B-token budget it is
 data-limited: fluent for a clause or two, near chance on knowledge benchmarks. The point is
@@ -43,7 +43,7 @@ run on owned hardware.
 - **Ternary weights `{−1, 0, +1}`** on all internal projections.
 - **Clean-room** architecture and ternary training operator.
-- **Lossless 182 MB GGUF**, **CPU-only** inference.
 - **Trained from scratch for ≈$150** on a single RTX A6000.
 ### Model Variants

 **Micro-Terse** is a 423M-parameter (≈320M active) **ternary-weight** language model trained from
 scratch for ≈**$150**, deployable as a **182 MB CPU-only GGUF**. Its weights are constrained to
+`{−1, 0, +1}` (≈1.58 bits), so `TQ2_0` packs them exactly; the released 182 MB file pairs that with a Q6_K tied embedding.
 It is a research proof-of-concept, **not** a production assistant. At an 8B-token budget it is
 data-limited: fluent for a clause or two, near chance on knowledge benchmarks. The point is
 - **Ternary weights `{−1, 0, +1}`** on all internal projections.
 - **Clean-room** architecture and ternary training operator.
+- **182 MB GGUF** (ternary weights packed exactly; Q6_K tied embedding), **CPU-only** inference.
 - **Trained from scratch for ≈$150** on a single RTX A6000.
 ### Model Variants