Instructions to use Akahsizrr/Mini-Whale-1-12B with libraries, inference providers, notebooks, and local apps. Follow these links to get started.

Libraries

How to use Akahsizrr/Mini-Whale-1-12B with Transformers:

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

pipe = pipeline("text-generation", model="Akahsizrr/Mini-Whale-1-12B", trust_remote_code=True)
messages = [
    {"role": "user", "content": "Who are you?"},
]
pipe(messages)

# Load model directly
from transformers import AutoModelForCausalLM
model = AutoModelForCausalLM.from_pretrained("Akahsizrr/Mini-Whale-1-12B", trust_remote_code=True, device_map="auto")

Notebooks
Google Colab
Kaggle
Local Apps Settings

vLLM

How to use Akahsizrr/Mini-Whale-1-12B with vLLM:

Install from pip and serve model

# Install vLLM from pip:
pip install vllm
# Start the vLLM server:
vllm serve "Akahsizrr/Mini-Whale-1-12B"
# Call the server using curl (OpenAI-compatible API):
curl -X POST "http://localhost:8000/v1/chat/completions" \
	-H "Content-Type: application/json" \
	--data '{
		"model": "Akahsizrr/Mini-Whale-1-12B",
		"messages": [
			{
				"role": "user",
				"content": "What is the capital of France?"
			}
		]
	}'

Use Docker

docker model run hf.co/Akahsizrr/Mini-Whale-1-12B

SGLang

How to use Akahsizrr/Mini-Whale-1-12B 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 "Akahsizrr/Mini-Whale-1-12B" \
    --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": "Akahsizrr/Mini-Whale-1-12B",
		"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 "Akahsizrr/Mini-Whale-1-12B" \
        --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": "Akahsizrr/Mini-Whale-1-12B",
		"messages": [
			{
				"role": "user",
				"content": "What is the capital of France?"
			}
		]
	}'

Docker Model Runner
How to use Akahsizrr/Mini-Whale-1-12B with Docker Model Runner:
```
docker model run hf.co/Akahsizrr/Mini-Whale-1-12B
```

are we there yet?

by pirola - opened 5 days ago

Discussion

pirola

5 days ago

are we there yet?

Akahsizrr

Owner 5 days ago

https://huggingface.co/Akahsizrr/Mini-Whale-1-12B HERE WE GO! with a few issues though, speed is not the best will update it tomorrow, and mlx, vllm and all the other ones arent availalbe yet, will add them tomorrow.
might build my own version of llama and mlx just for my architecture ahahah.

new better checkpoint coming tomorrow too, im all in on this model for the next week

Mindweller

5 days ago

Nice! Sir, what is the reason You chose QWEN3-4B and not LFM2.5-2.6B ?

Akahsizrr

Owner 5 days ago

I started this model before fuse-1, I don't think LFM released their 2.6B model when i started working on this

pirola

4 days ago

bro that's too slow.

pirola

4 days ago

What's the advantage that you reached?

Akahsizrr

Owner 4 days ago

its very slow but if i continue to work on it it has the potential to basically be deepseek v4 flash level intelligence, local, and at over 50 tps if i continue to lock in.

i get it at 10 tps on my rtx 3060. 2 days ago it was at 0.06 tps. this is real progression day after day.

Today I finished 1500 steps of training on the model, making the deepseek weights better inside the model too. 1-2 more models like this and i'll be able to provide you guys with local FRONTIER

pirola

about 23 hours ago

that's a strong statement that can only be sustained through benchmarks

Upload images, audio, and videos by dragging in the text input, pasting, or clicking here.

Tap or paste here to upload images

· Sign up or log in to comment