Instructions to use LetheanNetwork/lemer-bk with libraries, inference providers, notebooks, and local apps. Follow these links to get started.

Libraries

How to use LetheanNetwork/lemer-bk with Transformers:

# Load model directly
from transformers import AutoProcessor, AutoModelForMultimodalLM

processor = AutoProcessor.from_pretrained("LetheanNetwork/lemer-bk")
model = AutoModelForMultimodalLM.from_pretrained("LetheanNetwork/lemer-bk", device_map="auto")
messages = [
    {
        "role": "user",
        "content": [
            {"type": "image", "url": "https://huggingface.co/datasets/huggingface/documentation-images/resolve/main/p-blog/candy.JPG"},
            {"type": "text", "text": "What animal is on the candy?"}
        ]
    },
]
inputs = processor.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(processor.decode(outputs[0][inputs["input_ids"].shape[-1]:]))

Notebooks
Google Colab
Kaggle
Local Apps Settings

llama.cpp

How to use LetheanNetwork/lemer-bk 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 LetheanNetwork/lemer-bk:Q4_K_M
# Run inference directly in the terminal:
llama cli -hf LetheanNetwork/lemer-bk:Q4_K_M

Install from WinGet (Windows)

winget install llama.cpp
# Start a local OpenAI-compatible server with a web UI:
llama serve -hf LetheanNetwork/lemer-bk:Q4_K_M
# Run inference directly in the terminal:
llama cli -hf LetheanNetwork/lemer-bk:Q4_K_M

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 LetheanNetwork/lemer-bk:Q4_K_M
# Run inference directly in the terminal:
./llama-cli -hf LetheanNetwork/lemer-bk:Q4_K_M

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 LetheanNetwork/lemer-bk:Q4_K_M
# Run inference directly in the terminal:
./build/bin/llama-cli -hf LetheanNetwork/lemer-bk:Q4_K_M

Use Docker

docker model run hf.co/LetheanNetwork/lemer-bk:Q4_K_M

LM Studio
Jan
Ollama
How to use LetheanNetwork/lemer-bk with Ollama:
```
ollama run hf.co/LetheanNetwork/lemer-bk:Q4_K_M
```

Unsloth Studio

How to use LetheanNetwork/lemer-bk 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 LetheanNetwork/lemer-bk 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 LetheanNetwork/lemer-bk to start chatting

Using HuggingFace Spaces for Unsloth

# No setup required
# Open https://huggingface.co/spaces/unsloth/studio in your browser
# Search for LetheanNetwork/lemer-bk to start chatting

How to use LetheanNetwork/lemer-bk with Pi:

Start the llama.cpp server

# Install llama.cpp:
brew install llama.cpp
# Start a local OpenAI-compatible server:
llama serve -hf LetheanNetwork/lemer-bk:Q4_K_M

Configure the model in Pi

# Install Pi:
npm install -g @mariozechner/pi-coding-agent
# Add to ~/.pi/agent/models.json:
{
  "providers": {
    "llama-cpp": {
      "baseUrl": "http://localhost:8080/v1",
      "api": "openai-completions",
      "apiKey": "none",
      "models": [
        {
          "id": "LetheanNetwork/lemer-bk:Q4_K_M"
        }
      ]
    }
  }
}

Run Pi

# Start Pi in your project directory:
pi

Docker Model Runner
How to use LetheanNetwork/lemer-bk with Docker Model Runner:
```
docker model run hf.co/LetheanNetwork/lemer-bk:Q4_K_M
```

Lemonade

How to use LetheanNetwork/lemer-bk with Lemonade:

Pull the model

# Download Lemonade from https://lemonade-server.ai/
lemonade pull LetheanNetwork/lemer-bk:Q4_K_M

Run and chat with the model

lemonade run user.lemer-bk-Q4_K_M

List all available models

lemonade list

Hermes Agent

How to use LetheanNetwork/lemer-bk with Hermes Agent:

Start the llama.cpp server

# Install llama.cpp:
brew install llama.cpp
# Start a local OpenAI-compatible server:
llama serve -hf LetheanNetwork/lemer-bk:Q4_K_M

Configure Hermes

# Install Hermes:
curl -fsSL https://hermes-agent.nousresearch.com/install.sh | bash
hermes setup
# Point Hermes at the local server:
hermes config set model.provider custom
hermes config set model.base_url http://127.0.0.1:8080/v1
hermes config set model.default LetheanNetwork/lemer-bk:Q4_K_M

Run Hermes

hermes

Atomic Chat

OpenClaw

How to use LetheanNetwork/lemer-bk with OpenClaw:

Start the llama.cpp server

# Install llama.cpp:
brew install llama.cpp
# Start a local OpenAI-compatible server:
llama serve -hf LetheanNetwork/lemer-bk:Q4_K_M

Configure OpenClaw

# Install OpenClaw:
npm install -g openclaw@latest
# Register the local server and set it as the default model:
openclaw onboard --non-interactive --mode local \
  --auth-choice custom-api-key \
  --custom-base-url http://127.0.0.1:8080/v1 \
  --custom-model-id "LetheanNetwork/lemer-bk:Q4_K_M" \
  --custom-provider-id llama-cpp \
  --custom-compatibility openai \
  --custom-text-input \
  --accept-risk \
  --skip-health

Run OpenClaw

openclaw agent --local --agent main --message "Hello from Hugging Face"

Snider Virgil commited on Apr 6

Commit

cc4b6e4

1 Parent(s): 3a1587c

docs: add MMLU-Pro best-per-category table (46.4%), 5-shot results

Browse files

Math 80%, CS 65%, Biology 65% (0-shot). Full methodology with footnotes.

Co-Authored-By: Virgil <virgil@lethean.io>

Files changed (2) hide show

README.md +29 -5
results/lemer-bf16-all-5shot-multiturn-think-temp1.json +0 -0

README.md CHANGED Viewed

@@ -23,7 +23,35 @@ A [Gemma 4 E2B](https://huggingface.co/google/gemma-4-E2B-it) finetune by [lthn.
 ## Benchmarks
-### Lemer vs Stock Gemma 4 E2B (bf16)
 Columns: **(Think, Temperature)** — `G4` = Stock Gemma 4 E2B, `Lemer` = LEK-activated
@@ -45,10 +73,6 @@ Columns: **(Think, Temperature)** — `G4` = Stock Gemma 4 E2B, `Lemer` = LEK-ac
 | Psychology | TBC | TBC | TBC | TBC | TBC | 25.0% | TBC | TBC |
 | **Average** | TBC | TBC | TBC | TBC | TBC | **36.8%** | TBC | TBC |
-MMLU-Pro ([TIGER-Lab/MMLU-Pro](https://huggingface.co/datasets/TIGER-Lab/MMLU-Pro), test split, 20 samples per category, 5-shot CoT multi-turn).
-Evaluated using [rapid-mlx](https://github.com/LetheanNetwork/Rapid-MLX) + [OpenAI SDK](https://github.com/openai/openai-python) + Google [parse_response()](https://huggingface.co/google/gemma-4-E2B-it).
-Also verified via [mlx_lm](https://github.com/ml-explore/mlx-lm) native inference.
 ### Lemer Quantisation Benchmarks (MMLU-Pro, all categories)
 |  | [bf16](https://huggingface.co/lthn/lemer/tree/bf16) | [8bit](https://huggingface.co/lthn/lemer/tree/8bit) | [6bit](https://huggingface.co/lthn/lemer/tree/6bit) | [5bit](https://huggingface.co/lthn/lemer/tree/5bit) | [4bit](https://huggingface.co/lthn/lemer/tree/4bit) | [mxfp8](https://huggingface.co/lthn/lemer/tree/mxfp8) | [mxfp4](https://huggingface.co/lthn/lemer/tree/mxfp4) | [nvfp4](https://huggingface.co/lthn/lemer/tree/nvfp4) |

 ## Benchmarks
+### MMLU-Pro (bf16, best per category)
+|  | Lemer | Method |
+| :---- | :----: | :----: |
+| Math | **80.0%** | [5] |
+| Computer Science | **65.0%** | [5] |
+| Biology | **65.0%** | [0] |
+| Engineering | **55.0%** | [5] |
+| Other | **55.0%** | [5] |
+| Business | **50.0%** | [5] |
+| Physics | **50.0%** | [0] |
+| Economics | **45.0%** | [5] |
+| Psychology | **45.0%** | [5] |
+| Chemistry | **35.0%** | [5] |
+| Health | **35.0%** | [5] |
+| Philosophy | **30.0%** | [5] |
+| Law | **25.0%** | [0] |
+| History | **15.0%** | [5] |
+| **Average** | **46.4%** | |
+1. `[0]` 0-shot, think=on, temp=1.0
+2. `[5]` 5-shot CoT multi-turn, think=on, temp=1.0
+[TIGER-Lab/MMLU-Pro](https://huggingface.co/datasets/TIGER-Lab/MMLU-Pro) test split, 20 samples per category.
+Evaluated using [rapid-mlx](https://github.com/LetheanNetwork/Rapid-MLX) + [OpenAI SDK](https://github.com/openai/openai-python) + Google [parse_response()](https://huggingface.co/google/gemma-4-E2B-it).
+Verified via [mlx_lm](https://github.com/ml-explore/mlx-lm) native inference.
+Source: [eval.py](https://github.com/LetheanNetwork/LEM/blob/main/eval.py)
+### Lemer vs Stock Gemma 4 E2B (bf16, 0-shot)
 Columns: **(Think, Temperature)** — `G4` = Stock Gemma 4 E2B, `Lemer` = LEK-activated
 | Psychology | TBC | TBC | TBC | TBC | TBC | 25.0% | TBC | TBC |
 | **Average** | TBC | TBC | TBC | TBC | TBC | **36.8%** | TBC | TBC |
 ### Lemer Quantisation Benchmarks (MMLU-Pro, all categories)
 |  | [bf16](https://huggingface.co/lthn/lemer/tree/bf16) | [8bit](https://huggingface.co/lthn/lemer/tree/8bit) | [6bit](https://huggingface.co/lthn/lemer/tree/6bit) | [5bit](https://huggingface.co/lthn/lemer/tree/5bit) | [4bit](https://huggingface.co/lthn/lemer/tree/4bit) | [mxfp8](https://huggingface.co/lthn/lemer/tree/mxfp8) | [mxfp4](https://huggingface.co/lthn/lemer/tree/mxfp4) | [nvfp4](https://huggingface.co/lthn/lemer/tree/nvfp4) |

results/lemer-bf16-all-5shot-multiturn-think-temp1.json ADDED Viewed

The diff for this file is too large to render. See raw diff