feat: add scaleway inference provider

[Gnoale]

Hello!
I tried locally a few methods (feature_extraction and chat_completion) and it works perfectly

[HuggingFaceDocBuilderDev]

The docs for this PR live here. All of your documentation changes will be reflected on that endpoint. The docs are available until 30 days after the last update.

[Wauplin]

Looks great! I left minor comments regarding definition order but that's all.

I couldn't test the feature extraction model as I didn't know which model to pick.

I tested the conversational snippet by running this script locally and it worked as expected:

from huggingface_hub import InferenceClient

client = InferenceClient(provider="scaleway")
completion = client.chat.completions.create(
    model="meta-llama/Llama-3.1-8B-Instruct",
    messages=[{"role": "user", "content": "What is the capital of France?"}],
)

print(completion.choices[0].message.content)

Model meta-llama/Llama-3.1-8B-Instruct is in staging mode for provider scaleway. Meant for test purposes only.
I'm not aware of any tool call or library provided by the user. I can tell you directly that the capital of France is Paris.

[Gnoale]

Looks great! I left minor comments regarding definition order but that's all.

I couldn't test the feature extraction model as I didn't know which model to pick.

I tested the conversational snippet by running this script locally and it worked as expected:

from huggingface_hub import InferenceClient

client = InferenceClient(provider="scaleway")
completion = client.chat.completions.create(
    model="meta-llama/Llama-3.1-8B-Instruct",
    messages=[{"role": "user", "content": "What is the capital of France?"}],
)

print(completion.choices[0].message.content)

Model meta-llama/Llama-3.1-8B-Instruct is in staging mode for provider scaleway. Meant for test purposes only.
I'm not aware of any tool call or library provided by the user. I can tell you directly that the capital of France is Paris.

Cool ! ok perfect I updated the order

For feature extraction you can use BAAI/bge-multilingual-gemma2

[Wauplin]

Cool ! ok perfect I updated the order
For feature extraction you can use BAAI/bge-multilingual-gemma2

Thanks! Can confirm it works for me:

from huggingface_hub import InferenceClient

client = InferenceClient(provider="scaleway")

result = client.feature_extraction(
    "Today is a sunny day and I will get some ice cream.",
    model="BAAI/bge-multilingual-gemma2",
)

print(repr(result))

Model BAAI/bge-multilingual-gemma2 is in staging mode for provider scaleway. Meant for test purposes only.
array([[ 0.00822449,  0.00647736,  0.05108643, ..., -0.00762177,
        -0.0141983 ,  0.00327492]], shape=(1, 3584), dtype=float32)

[Wauplin]

Thanks @Gnoale! Looks good on my side 🤗

Will keep it open if @hanouticelina @SBrandeis want to have a quick look but otherwise we should be good to merge :)

[hanouticelina]

thanks @Gnoale for the contribution! 🤗