[InferenceSnippet] Take token from env variable if not set

[Wauplin]

Solve #1361.

Long awaited feature for @gary149. I did not go for the cleanest solution but it works well and should be robust/flexible enough if we need to fix something in the future.

EDIT: breaking change => access token should be passed as opts.accessToken now in snippets.getInferenceSnippets

TODO

once merged:

• adapt in moon-landing for snippets on model page https://github.com/huggingface-internal/moon-landing/pull/13964
• adapt in doc-builder for <inferencesnippet> html tag (used in hub-docs) Access token from env in snippets doc-builder#570
• hardcoded examples in hub-docs replace hf_xxxxxxxxxxxx by process.env.HF_TOKEN in examples hub-docs#1764

Some examples:

JS client

import { InferenceClient } from "@huggingface/inference";

const client = new InferenceClient(process.env.HF_TOKEN);

const chatCompletion = await client.chatCompletion({
    provider: "hf-inference",
    model: "meta-llama/Llama-3.1-8B-Instruct",
    messages: [
        {
            role: "user",
            content: "What is the capital of France?",
        },
    ],
});

console.log(chatCompletion.choices[0].message);

Python client

import os
from huggingface_hub import InferenceClient

client = InferenceClient(
    provider="hf-inference",
    api_key=os.environ["HF_TOKEN"],
)

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)

openai client

import os
from openai import OpenAI

client = OpenAI(
    base_url="https://router.huggingface.co/hf-inference/models/meta-llama/Llama-3.1-8B-Instruct/v1",
    api_key=os.environ["HF_TOKEN"],
)

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)

curl

curl https://router.huggingface.co/hf-inference/models/meta-llama/Llama-3.1-8B-Instruct/v1/chat/completions \
    -H "Authorization: Bearer $HF_TOKEN" \
    -H 'Content-Type: application/json' \
    -d '{
        "messages": [
            {
                "role": "user",
                "content": "What is the capital of France?"
            }
        ],
        "model": "meta-llama/Llama-3.1-8B-Instruct",
        "stream": false
    }'

check out PR diff for more examples

[Wauplin]

not related to this PR but fixes lint in PR (introduced in #1503)

[SBrandeis]

Excellent!

[hanouticelina]

Reviewed the snippets, looks good to me!

[SBrandeis]

Merging

[julien-c]

nice!