[draft] Support providerModelId in inference snippets (#1210)

Goal of this PR is to correctly handle model id mapping when generating
inference snippets for a given provider.

For now it's a simple PR to showcase what I had in mind in
huggingface-internal/moon-landing#12626 (comment)
(private repo). I only implemented it for chat-completion curl snippet
but the rest would follow the same pattern.

Note that we should use the mapping "only" for curl, pure Python, pure
JS and openai client snippets. For `huggingface.js` and
`huggingface_hub` ones, the model id from the Hub should be used.


---

**Note:** orthogonal to this PR but I realized that some URLs are also
incorrect depending on the provider.

---------

Co-authored-by: SBrandeis <[email protected]>

Author: Wauplin

packages/tasks-gen/scripts/generate-snippets-fixtures.ts

@@ -52,7 +52,7 @@ const TEST_CASES: {
 			inference: "",
 		},
 		languages: ["sh", "js", "py"],
-		providers: ["hf-inference"],
+		providers: ["hf-inference", "together"],
 		opts: { streaming: true },
 	},
 	{
@@ -64,7 +64,7 @@ const TEST_CASES: {
 			inference: "",
 		},
 		languages: ["sh", "js", "py"],
-		providers: ["hf-inference"],
+		providers: ["hf-inference", "fireworks-ai"],
 		opts: { streaming: false },
 	},
 	{
@@ -76,7 +76,7 @@ const TEST_CASES: {
 			inference: "",
 		},
 		languages: ["sh", "js", "py"],
-		providers: ["hf-inference"],
+		providers: ["hf-inference", "fireworks-ai"],
 		opts: { streaming: true },
 	},
 	{
@@ -87,7 +87,7 @@ const TEST_CASES: {
 			tags: [],
 			inference: "",
 		},
-		providers: ["hf-inference"],
+		providers: ["hf-inference", "fal-ai"],
 		languages: ["sh", "js", "py"],
 	},
 	{
@@ -133,7 +133,8 @@ function generateInferenceSnippet(
 	provider: SnippetInferenceProvider,
 	opts?: Record<string, unknown>
 ): InferenceSnippet[] {
-	return GET_SNIPPET_FN[language](model, "api_token", provider, opts);
+	const providerModelId = provider === "hf-inference" ? model.id : `<${provider} alias for ${model.id}>`;
+	return GET_SNIPPET_FN[language](model, "api_token", provider, providerModelId, opts);
 }
 
 async function getExpectedInferenceSnippet(

packages/tasks-gen/snippets-fixtures/conversational-llm-non-stream/0.curl.together.sh

@@ -2,7 +2,7 @@ curl 'https://router.huggingface.co/together/v1/chat/completions' \
 -H 'Authorization: Bearer api_token' \
 -H 'Content-Type: application/json' \
 --data '{
-    "model": "meta-llama/Llama-3.1-8B-Instruct",
+    "model": "<together alias for meta-llama/Llama-3.1-8B-Instruct>",
     "messages": [
 		{
 			"role": "user",

packages/tasks-gen/snippets-fixtures/conversational-llm-non-stream/1.openai.together.js

@@ -6,7 +6,7 @@ const client = new OpenAI({
 });
 
 const chatCompletion = await client.chat.completions.create({
-	model: "meta-llama/Llama-3.1-8B-Instruct",
+	model: "<together alias for meta-llama/Llama-3.1-8B-Instruct>",
 	messages: [
 		{
 			role: "user",

packages/tasks-gen/snippets-fixtures/conversational-llm-non-stream/1.openai.together.py

@@ -13,7 +13,7 @@
 ]
 
 completion = client.chat.completions.create(
-	model="meta-llama/Llama-3.1-8B-Instruct", 
+	model="<together alias for meta-llama/Llama-3.1-8B-Instruct>", 
 	messages=messages, 
 	max_tokens=500,
 )

packages/tasks-gen/snippets-fixtures/conversational-llm-stream/0.curl.together.sh

@@ -0,0 +1,14 @@
+curl 'https://router.huggingface.co/together/v1/chat/completions' \
+-H 'Authorization: Bearer api_token' \
+-H 'Content-Type: application/json' \
+--data '{
+    "model": "<together alias for meta-llama/Llama-3.1-8B-Instruct>",
+    "messages": [
+		{
+			"role": "user",
+			"content": "What is the capital of France?"
+		}
+	],
+    "max_tokens": 500,
+    "stream": true
+}'

packages/tasks-gen/snippets-fixtures/conversational-llm-stream/0.huggingface.js.together.js

@@ -0,0 +1,25 @@
+import { HfInference } from "@huggingface/inference";
+
+const client = new HfInference("api_token");
+
+let out = "";
+
+const stream = client.chatCompletionStream({
+	model: "meta-llama/Llama-3.1-8B-Instruct",
+	messages: [
+		{
+			role: "user",
+			content: "What is the capital of France?"
+		}
+	],
+	provider: "together",
+	max_tokens: 500,
+});
+
+for await (const chunk of stream) {
+	if (chunk.choices && chunk.choices.length > 0) {
+		const newContent = chunk.choices[0].delta.content;
+		out += newContent;
+		console.log(newContent);
+	}  
+}

packages/tasks-gen/snippets-fixtures/conversational-llm-stream/0.huggingface_hub.together.py

@@ -0,0 +1,23 @@
+from huggingface_hub import InferenceClient
+
+client = InferenceClient(
+	provider="together",
+	api_key="api_token"
+)
+
+messages = [
+	{
+		"role": "user",
+		"content": "What is the capital of France?"
+	}
+]
+
+stream = client.chat.completions.create(
+	model="meta-llama/Llama-3.1-8B-Instruct", 
+	messages=messages, 
+	max_tokens=500,
+	stream=True
+)
+
+for chunk in stream:
+    print(chunk.choices[0].delta.content, end="")

packages/tasks-gen/snippets-fixtures/conversational-llm-stream/1.openai.together.js

@@ -0,0 +1,28 @@
+import { OpenAI } from "openai";
+
+const client = new OpenAI({
+	baseURL: "https://router.huggingface.co/together",
+	apiKey: "api_token"
+});
+
+let out = "";
+
+const stream = await client.chat.completions.create({
+	model: "<together alias for meta-llama/Llama-3.1-8B-Instruct>",
+	messages: [
+		{
+			role: "user",
+			content: "What is the capital of France?"
+		}
+	],
+	max_tokens: 500,
+	stream: true,
+});
+
+for await (const chunk of stream) {
+	if (chunk.choices && chunk.choices.length > 0) {
+		const newContent = chunk.choices[0].delta.content;
+		out += newContent;
+		console.log(newContent);
+	}  
+}

packages/tasks-gen/snippets-fixtures/conversational-llm-stream/1.openai.together.py

@@ -0,0 +1,23 @@
+from openai import OpenAI
+
+client = OpenAI(
+	base_url="https://router.huggingface.co/together",
+	api_key="api_token"
+)
+
+messages = [
+	{
+		"role": "user",
+		"content": "What is the capital of France?"
+	}
+]
+
+stream = client.chat.completions.create(
+    model="<together alias for meta-llama/Llama-3.1-8B-Instruct>", 
+	messages=messages, 
+	max_tokens=500,
+	stream=True
+)
+
+for chunk in stream:
+	print(chunk.choices[0].delta.content, end="")

packages/tasks-gen/snippets-fixtures/conversational-vlm-non-stream/0.curl.fireworks-ai.sh

@@ -0,0 +1,25 @@
+curl 'https://router.huggingface.co/fireworks-ai/v1/chat/completions' \
+-H 'Authorization: Bearer api_token' \
+-H 'Content-Type: application/json' \
+--data '{
+    "model": "<fireworks-ai alias for meta-llama/Llama-3.2-11B-Vision-Instruct>",
+    "messages": [
+		{
+			"role": "user",
+			"content": [
+				{
+					"type": "text",
+					"text": "Describe this image in one sentence."
+				},
+				{
+					"type": "image_url",
+					"image_url": {
+						"url": "https://cdn.britannica.com/61/93061-050-99147DCE/Statue-of-Liberty-Island-New-York-Bay.jpg"
+					}
+				}
+			]
+		}
+	],
+    "max_tokens": 500,
+    "stream": false
+}'