Instructions to use Lernex/Metis-1.6-Logos-Base with libraries, inference providers, notebooks, and local apps. Follow these links to get started.

Libraries

How to use Lernex/Metis-1.6-Logos-Base with Transformers:

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

pipe = pipeline("text-generation", model="Lernex/Metis-1.6-Logos-Base", trust_remote_code=True)

# pip install -U transformers accelerate
# Load model directly
from transformers import AutoModelForCausalLM
model = AutoModelForCausalLM.from_pretrained("Lernex/Metis-1.6-Logos-Base", trust_remote_code=True, device_map="auto")

Notebooks
Google Colab
Kaggle
Local Apps Settings

vLLM

How to use Lernex/Metis-1.6-Logos-Base with vLLM:

Install from pip and serve model

# Install vLLM from pip:
pip install vllm
# Start the vLLM server:
vllm serve "Lernex/Metis-1.6-Logos-Base"
# Call the server using curl (OpenAI-compatible API):
curl -X POST "http://localhost:8000/v1/completions" \
	-H "Content-Type: application/json" \
	--data '{
		"model": "Lernex/Metis-1.6-Logos-Base",
		"prompt": "Once upon a time,",
		"max_tokens": 512,
		"temperature": 0.5
	}'

Use Docker

docker model run hf.co/Lernex/Metis-1.6-Logos-Base

SGLang

How to use Lernex/Metis-1.6-Logos-Base 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 "Lernex/Metis-1.6-Logos-Base" \
    --host 0.0.0.0 \
    --port 30000
# Call the server using curl (OpenAI-compatible API):
curl -X POST "http://localhost:30000/v1/completions" \
	-H "Content-Type: application/json" \
	--data '{
		"model": "Lernex/Metis-1.6-Logos-Base",
		"prompt": "Once upon a time,",
		"max_tokens": 512,
		"temperature": 0.5
	}'

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 "Lernex/Metis-1.6-Logos-Base" \
        --host 0.0.0.0 \
        --port 30000
# Call the server using curl (OpenAI-compatible API):
curl -X POST "http://localhost:30000/v1/completions" \
	-H "Content-Type: application/json" \
	--data '{
		"model": "Lernex/Metis-1.6-Logos-Base",
		"prompt": "Once upon a time,",
		"max_tokens": 512,
		"temperature": 0.5
	}'

Docker Model Runner
How to use Lernex/Metis-1.6-Logos-Base with Docker Model Runner:
```
docker model run hf.co/Lernex/Metis-1.6-Logos-Base
```

GiuliannoV commited on 7 days ago

Commit

bf2e47b

verified ·

1 Parent(s): 6a17736

Update benchmark_suite_4k.json with latest base model evaluation scores

Browse files

Files changed (1) hide show

benchmark_suite_4k.json +5 -5

benchmark_suite_4k.json CHANGED Viewed

@@ -170,7 +170,7 @@
  {
   "section": "Frontier reasoning and knowledge",
   "benchmark": "BIG-Bench Extra Hard",
-  "score": "pending",
   "protocol": "The benchmark's own prompt and scorer. BIG-Bench Extra Hard 0-shot, items too long for the window counted wrong. ARC-AGI-2 as text grids, two attempts. SciCode without background, the reply starting at the step's function header.",
   "detail": null,
   "others": {
@@ -338,7 +338,7 @@
  {
   "section": "Math",
   "benchmark": "Omni-MATH",
-  "score": "pending",
   "protocol": "Minerva 4-shot prompt, greedy. MATH-500 by exact match with a 256-token cap; AIME, HMMT and Omni-MATH checked by math_verify, Omni-MATH without its judge model.",
   "detail": null,
   "others": {
@@ -689,7 +689,7 @@
    "Granite 4.1 8B": "83.08",
    "Ministral 3 8B": "91.57",
    "Olmo 3 7B": "75.61",
-   "Falcon-H1 7B": "pending",
    "Qwen3.5 4B": "65.65",
    "Qwen3 4B": "70.17",
    "Gemma 4 E4B": "74.40"
@@ -710,7 +710,7 @@
    "Qwen3.5 9B": "52.31",
    "Qwen3 8B": "66.10",
    "Gemma 4 12B": "pending",
-   "Granite 4.1 8B": "pending",
    "Ministral 3 8B": "80.74",
    "Olmo 3 7B": "70.95",
    "Falcon-H1 7B": "pending",
@@ -1043,7 +1043,7 @@
    "Granite 4.0 H Tiny": "62.90",
    "Trinity Nano": "58.66",
    "OLMoE 1B-7B": "50.87",
-   "Qwen3.5 9B": "0.01",
    "Qwen3 8B": "58.96",
    "Gemma 4 12B": "63.08",
    "Granite 4.1 8B": "71.68",

  {
   "section": "Frontier reasoning and knowledge",
   "benchmark": "BIG-Bench Extra Hard",
+  "score": "4.60",
   "protocol": "The benchmark's own prompt and scorer. BIG-Bench Extra Hard 0-shot, items too long for the window counted wrong. ARC-AGI-2 as text grids, two attempts. SciCode without background, the reply starting at the step's function header.",
   "detail": null,
   "others": {
  {
   "section": "Math",
   "benchmark": "Omni-MATH",
+  "score": "6.93",
   "protocol": "Minerva 4-shot prompt, greedy. MATH-500 by exact match with a 256-token cap; AIME, HMMT and Omni-MATH checked by math_verify, Omni-MATH without its judge model.",
   "detail": null,
   "others": {
    "Granite 4.1 8B": "83.08",
    "Ministral 3 8B": "91.57",
    "Olmo 3 7B": "75.61",
+   "Falcon-H1 7B": "86.25",
    "Qwen3.5 4B": "65.65",
    "Qwen3 4B": "70.17",
    "Gemma 4 E4B": "74.40"
    "Qwen3.5 9B": "52.31",
    "Qwen3 8B": "66.10",
    "Gemma 4 12B": "pending",
+   "Granite 4.1 8B": "78.04",
    "Ministral 3 8B": "80.74",
    "Olmo 3 7B": "70.95",
    "Falcon-H1 7B": "pending",
    "Granite 4.0 H Tiny": "62.90",
    "Trinity Nano": "58.66",
    "OLMoE 1B-7B": "50.87",
+   "Qwen3.5 9B": "pending",
    "Qwen3 8B": "58.96",
    "Gemma 4 12B": "63.08",
    "Granite 4.1 8B": "71.68",