Instructions to use webAI-Official/TwIL-LM3-Pro with libraries, inference providers, notebooks, and local apps. Follow these links to get started.

Libraries

How to use webAI-Official/TwIL-LM3-Pro with Transformers:

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

pipe = pipeline("text-generation", model="webAI-Official/TwIL-LM3-Pro")
messages = [
    {"role": "user", "content": "Who are you?"},
]
pipe(messages)

# Load model directly
from transformers import AutoTokenizer, AutoModelForCausalLM

tokenizer = AutoTokenizer.from_pretrained("webAI-Official/TwIL-LM3-Pro")
model = AutoModelForCausalLM.from_pretrained("webAI-Official/TwIL-LM3-Pro", device_map="auto")
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]:]))

Notebooks
Google Colab
Kaggle
Local Apps Settings

llama.cpp

How to use webAI-Official/TwIL-LM3-Pro 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 webAI-Official/TwIL-LM3-Pro:Q4_K_M
# Run inference directly in the terminal:
llama cli -hf webAI-Official/TwIL-LM3-Pro:Q4_K_M

Install from WinGet (Windows)

winget install llama.cpp
# Start a local OpenAI-compatible server with a web UI:
llama serve -hf webAI-Official/TwIL-LM3-Pro:Q4_K_M
# Run inference directly in the terminal:
llama cli -hf webAI-Official/TwIL-LM3-Pro: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 webAI-Official/TwIL-LM3-Pro:Q4_K_M
# Run inference directly in the terminal:
./llama-cli -hf webAI-Official/TwIL-LM3-Pro: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 webAI-Official/TwIL-LM3-Pro:Q4_K_M
# Run inference directly in the terminal:
./build/bin/llama-cli -hf webAI-Official/TwIL-LM3-Pro:Q4_K_M

Use Docker

docker model run hf.co/webAI-Official/TwIL-LM3-Pro:Q4_K_M

LM Studio
Jan

vLLM

How to use webAI-Official/TwIL-LM3-Pro with vLLM:

Install from pip and serve model

# Install vLLM from pip:
pip install vllm
# Start the vLLM server:
vllm serve "webAI-Official/TwIL-LM3-Pro"
# Call the server using curl (OpenAI-compatible API):
curl -X POST "http://localhost:8000/v1/chat/completions" \
	-H "Content-Type: application/json" \
	--data '{
		"model": "webAI-Official/TwIL-LM3-Pro",
		"messages": [
			{
				"role": "user",
				"content": "What is the capital of France?"
			}
		]
	}'

Use Docker

docker model run hf.co/webAI-Official/TwIL-LM3-Pro:Q4_K_M

SGLang

How to use webAI-Official/TwIL-LM3-Pro 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 "webAI-Official/TwIL-LM3-Pro" \
    --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": "webAI-Official/TwIL-LM3-Pro",
		"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 "webAI-Official/TwIL-LM3-Pro" \
        --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": "webAI-Official/TwIL-LM3-Pro",
		"messages": [
			{
				"role": "user",
				"content": "What is the capital of France?"
			}
		]
	}'

Ollama
How to use webAI-Official/TwIL-LM3-Pro with Ollama:
```
ollama run hf.co/webAI-Official/TwIL-LM3-Pro:Q4_K_M
```
Unsloth Desktop

How to use webAI-Official/TwIL-LM3-Pro with Pi:

Start the llama.cpp server

# Install llama.cpp:
brew install llama.cpp
# Start a local OpenAI-compatible server:
llama serve -hf webAI-Official/TwIL-LM3-Pro:Q4_K_M

Configure the model in Pi

# Install Pi:
npm install -g @earendil-works/pi-coding-agent
# Add to ~/.pi/agent/models.json:
{
  "providers": {
    "llama-cpp": {
      "baseUrl": "http://localhost:8080/v1",
      "api": "openai-completions",
      "apiKey": "none",
      "models": [
        {
          "id": "webAI-Official/TwIL-LM3-Pro:Q4_K_M"
        }
      ]
    }
  }
}

Run Pi

# Start Pi in your project directory:
pi

Docker Model Runner
How to use webAI-Official/TwIL-LM3-Pro with Docker Model Runner:
```
docker model run hf.co/webAI-Official/TwIL-LM3-Pro:Q4_K_M
```

Lemonade

How to use webAI-Official/TwIL-LM3-Pro with Lemonade:

Pull the model

# Download Lemonade from https://lemonade-server.ai/
lemonade pull webAI-Official/TwIL-LM3-Pro:Q4_K_M

Run and chat with the model

lemonade run user.TwIL-LM3-Pro-Q4_K_M

List all available models

lemonade list

Hermes Agent

How to use webAI-Official/TwIL-LM3-Pro with Hermes Agent:

Start the llama.cpp server

# Install llama.cpp:
brew install llama.cpp
# Start a local OpenAI-compatible server:
llama serve -hf webAI-Official/TwIL-LM3-Pro: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 webAI-Official/TwIL-LM3-Pro:Q4_K_M

Run Hermes

hermes

Atomic Chat

OpenClaw

How to use webAI-Official/TwIL-LM3-Pro with OpenClaw:

Start the llama.cpp server

# Install llama.cpp:
brew install llama.cpp
# Start a local OpenAI-compatible server:
llama serve -hf webAI-Official/TwIL-LM3-Pro: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 "webAI-Official/TwIL-LM3-Pro: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"

anurag051194 commited on 3 days ago

Commit

588354c

verified ·

1 Parent(s): a8247e4

VibeThinker-webAI-trained: add estimated tok/s

Browse files

Files changed (1) hide show

README.md +8 -3

README.md CHANGED Viewed

@@ -129,7 +129,7 @@ Cells marked † need the engine note below.
 | **macro gate** | **0.5539** | 0.4313 | 0.4118 | 0.508 | 0.4466 | 0.4218 | 0.2925 | 0.3473 | 0.3757 | 0.5336 | — |
 | **strict-7** | **0.2879** | 0.1821 | 0.2021 | — | 0.1121 | 0.1971 | 0.1229 | 0.1579 | 0.1714 | 0.2093 | — |
 | macro_primary | **0.5875** | 0.4825 | 0.4637 | 0.579 | 0.4313 | 0.4475 | 0.3450 | 0.4188 | 0.4213 | 0.5750 | — |
-| tok/s | 21169 † | 21916 † | 28010 † | — | 11097 † | 15880 | 16160 | 25230 | 22480 | 9420 | 3374 |
 | mean gen length | 1902 | 2951 | 2688 | — | 2486 | **564** | 696 | 2296 | 1830 | 2094 | 1005 |
 | **ans/s** | 11.1 † | 7.4 † | 10.4 † | — | 4.5 † | **28.1** | 23.2 | 10.9 | 12.0 | 4.5 | 3.4 |
@@ -153,7 +153,12 @@ family tables score `mcq_answer` with loose-match credit (0.695), so the strict
 six-lane average and strict-7 — which all need strict scoring — are left as —. Its perplexities
 come from that separate run; the family table records the untuned VibeThinker-3B at 18.70 and
 21.80 there, against 16.93 and 27.03 in the columns above, so read them within their own source.
-Throughput and generation length were not measured for it.
 † **Throughput for TwIL-LM3-Pro, its base and VibeThinker-3B** was measured with the same
 dedicated protocol and prompt file as the peer columns (128 prompts × 512 generated tokens, EOS
@@ -241,7 +246,7 @@ spots in absolute terms are `procedural` (strict 0.1200, loose 0.2350) and FOL t
 | math500 | 0.7467 | 0.6567 | 0.7900 | 0.790 | 0.3600 ¶¶¶ | 0.6900 | 0.4233 | 0.7133 | 0.7800 | 0.6100 | **0.8433** |
 | **macro (10 CoT datasets)** | 0.7901 | 0.7942 | 0.8097 | 0.802 | 0.7683 | 0.7339 | 0.6997 | 0.7523 | 0.7884 | 0.8493 | **0.8689** |
 | **macro (all 14)** | 0.7425 | 0.7332 | 0.7262 | 0.728 | 0.6611 | 0.6694 | 0.6245 | 0.6814 | 0.7378 | 0.7591 | **0.8086** |
-| tok/s | 21169 † | 21916 † | 28010 † | — | 11097 † | 15880 | 16160 | 25230 | 22480 | 9420 | 3374 |
 | mean gen length | ≈792 | ≈1282 | ≈1789 | — | ≈2787 | **482** | 510 | ≈796 | ≈1327 | ≈1931 | 801 |
 | **ans/s** | 26.7 † | 17.1 † | ≈15.7 † | — | ≈4.0 † | **32.9** | 31.7 | ≈31.7 | ≈16.9 | 4.9 | 4.2 |

 | **macro gate** | **0.5539** | 0.4313 | 0.4118 | 0.508 | 0.4466 | 0.4218 | 0.2925 | 0.3473 | 0.3757 | 0.5336 | — |
 | **strict-7** | **0.2879** | 0.1821 | 0.2021 | — | 0.1121 | 0.1971 | 0.1229 | 0.1579 | 0.1714 | 0.2093 | — |
 | macro_primary | **0.5875** | 0.4825 | 0.4637 | 0.579 | 0.4313 | 0.4475 | 0.3450 | 0.4188 | 0.4213 | 0.5750 | — |
+| tok/s | 21169 † | 21916 † | 28010 † | ≈28010 ★ † | 11097 † | 15880 | 16160 | 25230 | 22480 | 9420 | 3374 |
 | mean gen length | 1902 | 2951 | 2688 | — | 2486 | **564** | 696 | 2296 | 1830 | 2094 | 1005 |
 | **ans/s** | 11.1 † | 7.4 † | 10.4 † | — | 4.5 † | **28.1** | 23.2 | 10.9 | 12.0 | 4.5 | 3.4 |
 six-lane average and strict-7 — which all need strict scoring — are left as —. Its perplexities
 come from that separate run; the family table records the untuned VibeThinker-3B at 18.70 and
 21.80 there, against 16.93 and 27.03 in the columns above, so read them within their own source.
+Its `tok/s` is an **estimate, not a measurement**: merging changes weights but not architecture,
+parameter count or tokenizer, and the throughput protocol fixes the output at 512 generated tokens
+with EOS ignored, so the decode rate does not depend on what the weights say. It is therefore set
+equal to the 28,010 tok/s measured for the base VibeThinker-3B in the same session (a ≈ and both
+marks in the cell). Mean generation length, and so `ans/s`, does depend on the weights and was not
+measured, so those rows stay —.
 † **Throughput for TwIL-LM3-Pro, its base and VibeThinker-3B** was measured with the same
 dedicated protocol and prompt file as the peer columns (128 prompts × 512 generated tokens, EOS
 | math500 | 0.7467 | 0.6567 | 0.7900 | 0.790 | 0.3600 ¶¶¶ | 0.6900 | 0.4233 | 0.7133 | 0.7800 | 0.6100 | **0.8433** |
 | **macro (10 CoT datasets)** | 0.7901 | 0.7942 | 0.8097 | 0.802 | 0.7683 | 0.7339 | 0.6997 | 0.7523 | 0.7884 | 0.8493 | **0.8689** |
 | **macro (all 14)** | 0.7425 | 0.7332 | 0.7262 | 0.728 | 0.6611 | 0.6694 | 0.6245 | 0.6814 | 0.7378 | 0.7591 | **0.8086** |
+| tok/s | 21169 † | 21916 † | 28010 † | ≈28010 ★ † | 11097 † | 15880 | 16160 | 25230 | 22480 | 9420 | 3374 |
 | mean gen length | ≈792 | ≈1282 | ≈1789 | — | ≈2787 | **482** | 510 | ≈796 | ≈1327 | ≈1931 | 801 |
 | **ans/s** | 26.7 † | 17.1 † | ≈15.7 † | — | ≈4.0 † | **32.9** | 31.7 | ≈31.7 | ≈16.9 | 4.9 | 4.2 |