Skip to content

Commit 290314b

Browse files
hanouticelinahanouticelina
committed
Merge branch 'main' into regenerate-api-inference-docs
2 parents a914215 + 135c53e commit 290314b

File tree

6 files changed

+335
-100
lines changed

6 files changed

+335
-100
lines changed

‎docs/api-inference/tasks/chat-completion.md

+254-52
Original file line numberDiff line numberDiff line change
@@ -64,46 +64,133 @@ The API supports:
6464
curl 'https://api-inference.huggingface.co/models/google/gemma-2-2b-it/v1/chat/completions' \
6565
-H "Authorization: Bearer hf_***" \
6666
-H 'Content-Type: application/json' \
67-
-d '{
68-
"model": "google/gemma-2-2b-it",
69-
"messages": [{"role": "user", "content": "What is the capital of France?"}],
70-
"max_tokens": 500,
71-
"stream": false
67+
--data '{
68+
"model": "google/gemma-2-2b-it",
69+
"messages": [
70+
{
71+
"role": "user",
72+
"content": "What is the capital of France?"
73+
}
74+
],
75+
"max_tokens": 500,
76+
"stream": true
7277
}'
73-
7478
```
7579
</curl>
7680

7781
<python>
82+
With huggingface_hub client:
7883
```py
7984
from huggingface_hub import InferenceClient
8085

8186
client = InferenceClient(api_key="hf_***")
8287

83-
for message in client.chat_completion(
84-
model="google/gemma-2-2b-it",
85-
messages=[{"role": "user", "content": "What is the capital of France?"}],
88+
messages = [
89+
{
90+
"role": "user",
91+
"content": "What is the capital of France?"
92+
}
93+
]
94+
95+
stream = client.chat.completions.create(
96+
model="google/gemma-2-2b-it",
97+
messages=messages,
8698
max_tokens=500,
87-
stream=True,
88-
):
89-
print(message.choices[0].delta.content, end="")
99+
stream=True
100+
)
101+
102+
for chunk in stream:
103+
print(chunk.choices[0].delta.content, end="")
104+
```
105+
106+
With openai client:
107+
```py
108+
from openai import OpenAI
109+
110+
client = OpenAI(
111+
base_url="https://api-inference.huggingface.co/v1/",
112+
api_key="hf_***"
113+
)
114+
115+
messages = [
116+
{
117+
"role": "user",
118+
"content": "What is the capital of France?"
119+
}
120+
]
121+
122+
stream = client.chat.completions.create(
123+
model="google/gemma-2-2b-it",
124+
messages=messages,
125+
max_tokens=500,
126+
stream=True
127+
)
128+
129+
for chunk in stream:
130+
print(chunk.choices[0].delta.content, end="")
90131
```
91132

92133
To use the Python client, see `huggingface_hub`'s [package reference](https://huggingface.co/docs/huggingface_hub/package_reference/inference_client#huggingface_hub.InferenceClient.chat_completion).
93134
</python>
94135

95136
<js>
137+
With huggingface_hub client:
96138
```js
97-
import { HfInference } from "@huggingface/inference";
139+
import { HfInference } from "@huggingface/inference"
98140

99-
const inference = new HfInference("hf_***");
141+
const client = new HfInference("hf_***")
100142

101-
for await (const chunk of inference.chatCompletionStream({
143+
let out = "";
144+
145+
const stream = client.chatCompletionStream({
102146
model: "google/gemma-2-2b-it",
103-
messages: [{ role: "user", content: "What is the capital of France?" }],
147+
messages: [
148+
{
149+
role: "user",
150+
content: "What is the capital of France?"
151+
}
152+
],
153+
max_tokens: 500
154+
});
155+
156+
for await (const chunk of stream) {
157+
if (chunk.choices && chunk.choices.length > 0) {
158+
const newContent = chunk.choices[0].delta.content;
159+
out += newContent;
160+
console.log(newContent);
161+
}
162+
}
163+
```
164+
165+
With openai client:
166+
```js
167+
import { OpenAI } from "openai"
168+
169+
const client = new OpenAI({
170+
baseURL: "https://api-inference.huggingface.co/v1/",
171+
apiKey: "hf_***"
172+
})
173+
174+
let out = "";
175+
176+
const stream = await client.chat.completions.create({
177+
model: "google/gemma-2-2b-it",
178+
messages: [
179+
{
180+
role: "user",
181+
content: "What is the capital of France?"
182+
}
183+
],
104184
max_tokens: 500,
105-
})) {
106-
process.stdout.write(chunk.choices[0]?.delta?.content || "");
185+
stream: true,
186+
});
187+
188+
for await (const chunk of stream) {
189+
if (chunk.choices && chunk.choices.length > 0) {
190+
const newContent = chunk.choices[0].delta.content;
191+
out += newContent;
192+
console.log(newContent);
193+
}
107194
}
108195
```
109196

@@ -124,73 +211,188 @@ To use the JavaScript client, see `huggingface.js`'s [package reference](https:/
124211
curl 'https://api-inference.huggingface.co/models/meta-llama/Llama-3.2-11B-Vision-Instruct/v1/chat/completions' \
125212
-H "Authorization: Bearer hf_***" \
126213
-H 'Content-Type: application/json' \
127-
-d '{
128-
"model": "meta-llama/Llama-3.2-11B-Vision-Instruct",
129-
"messages": [
214+
--data '{
215+
"model": "meta-llama/Llama-3.2-11B-Vision-Instruct",
216+
"messages": [
130217
{
131218
"role": "user",
132219
"content": [
133-
{"type": "image_url", "image_url": {"url": "https://cdn.britannica.com/61/93061-050-99147DCE/Statue-of-Liberty-Island-New-York-Bay.jpg"}},
134-
{"type": "text", "text": "Describe this image in one sentence."}
220+
{
221+
"type": "text",
222+
"text": "Describe this image in one sentence."
223+
},
224+
{
225+
"type": "image_url",
226+
"image_url": {
227+
"url": "https://cdn.britannica.com/61/93061-050-99147DCE/Statue-of-Liberty-Island-New-York-Bay.jpg"
228+
}
229+
}
135230
]
136231
}
137232
],
138-
"max_tokens": 500,
139-
"stream": false
233+
"max_tokens": 500,
234+
"stream": true
140235
}'
141-
142236
```
143237
</curl>
144238

145239
<python>
240+
With huggingface_hub client:
146241
```py
147242
from huggingface_hub import InferenceClient
148243

149244
client = InferenceClient(api_key="hf_***")
150245

151-
image_url = "https://cdn.britannica.com/61/93061-050-99147DCE/Statue-of-Liberty-Island-New-York-Bay.jpg"
246+
messages = [
247+
{
248+
"role": "user",
249+
"content": [
250+
{
251+
"type": "text",
252+
"text": "Describe this image in one sentence."
253+
},
254+
{
255+
"type": "image_url",
256+
"image_url": {
257+
"url": "https://cdn.britannica.com/61/93061-050-99147DCE/Statue-of-Liberty-Island-New-York-Bay.jpg"
258+
}
259+
}
260+
]
261+
}
262+
]
263+
264+
stream = client.chat.completions.create(
265+
model="meta-llama/Llama-3.2-11B-Vision-Instruct",
266+
messages=messages,
267+
max_tokens=500,
268+
stream=True
269+
)
152270

153-
for message in client.chat_completion(
154-
model="meta-llama/Llama-3.2-11B-Vision-Instruct",
155-
messages=[
156-
{
157-
"role": "user",
158-
"content": [
159-
{"type": "image_url", "image_url": {"url": image_url}},
160-
{"type": "text", "text": "Describe this image in one sentence."},
161-
],
162-
}
163-
],
271+
for chunk in stream:
272+
print(chunk.choices[0].delta.content, end="")
273+
```
274+
275+
With openai client:
276+
```py
277+
from openai import OpenAI
278+
279+
client = OpenAI(
280+
base_url="https://api-inference.huggingface.co/v1/",
281+
api_key="hf_***"
282+
)
283+
284+
messages = [
285+
{
286+
"role": "user",
287+
"content": [
288+
{
289+
"type": "text",
290+
"text": "Describe this image in one sentence."
291+
},
292+
{
293+
"type": "image_url",
294+
"image_url": {
295+
"url": "https://cdn.britannica.com/61/93061-050-99147DCE/Statue-of-Liberty-Island-New-York-Bay.jpg"
296+
}
297+
}
298+
]
299+
}
300+
]
301+
302+
stream = client.chat.completions.create(
303+
model="meta-llama/Llama-3.2-11B-Vision-Instruct",
304+
messages=messages,
164305
max_tokens=500,
165-
stream=True,
166-
):
167-
print(message.choices[0].delta.content, end="")
306+
stream=True
307+
)
308+
309+
for chunk in stream:
310+
print(chunk.choices[0].delta.content, end="")
168311
```
169312

170313
To use the Python client, see `huggingface_hub`'s [package reference](https://huggingface.co/docs/huggingface_hub/package_reference/inference_client#huggingface_hub.InferenceClient.chat_completion).
171314
</python>
172315

173316
<js>
317+
With huggingface_hub client:
174318
```js
175-
import { HfInference } from "@huggingface/inference";
319+
import { HfInference } from "@huggingface/inference"
176320

177-
const inference = new HfInference("hf_***");
178-
const imageUrl = "https://cdn.britannica.com/61/93061-050-99147DCE/Statue-of-Liberty-Island-New-York-Bay.jpg";
321+
const client = new HfInference("hf_***")
179322

180-
for await (const chunk of inference.chatCompletionStream({
323+
let out = "";
324+
325+
const stream = client.chatCompletionStream({
181326
model: "meta-llama/Llama-3.2-11B-Vision-Instruct",
182327
messages: [
183328
{
184-
"role": "user",
185-
"content": [
186-
{"type": "image_url", "image_url": {"url": imageUrl}},
187-
{"type": "text", "text": "Describe this image in one sentence."},
188-
],
329+
role: "user",
330+
content: [
331+
{
332+
type: "text",
333+
text: "Describe this image in one sentence."
334+
},
335+
{
336+
type: "image_url",
337+
image_url: {
338+
url: "https://cdn.britannica.com/61/93061-050-99147DCE/Statue-of-Liberty-Island-New-York-Bay.jpg"
339+
}
340+
}
341+
]
342+
}
343+
],
344+
max_tokens: 500
345+
});
346+
347+
for await (const chunk of stream) {
348+
if (chunk.choices && chunk.choices.length > 0) {
349+
const newContent = chunk.choices[0].delta.content;
350+
out += newContent;
351+
console.log(newContent);
352+
}
353+
}
354+
```
355+
356+
With openai client:
357+
```js
358+
import { OpenAI } from "openai"
359+
360+
const client = new OpenAI({
361+
baseURL: "https://api-inference.huggingface.co/v1/",
362+
apiKey: "hf_***"
363+
})
364+
365+
let out = "";
366+
367+
const stream = await client.chat.completions.create({
368+
model: "meta-llama/Llama-3.2-11B-Vision-Instruct",
369+
messages: [
370+
{
371+
role: "user",
372+
content: [
373+
{
374+
type: "text",
375+
text: "Describe this image in one sentence."
376+
},
377+
{
378+
type: "image_url",
379+
image_url: {
380+
url: "https://cdn.britannica.com/61/93061-050-99147DCE/Statue-of-Liberty-Island-New-York-Bay.jpg"
381+
}
382+
}
383+
]
189384
}
190385
],
191386
max_tokens: 500,
192-
})) {
193-
process.stdout.write(chunk.choices[0]?.delta?.content || "");
387+
stream: true,
388+
});
389+
390+
for await (const chunk of stream) {
391+
if (chunk.choices && chunk.choices.length > 0) {
392+
const newContent = chunk.choices[0].delta.content;
393+
out += newContent;
394+
console.log(newContent);
395+
}
194396
}
195397
```
196398

‎docs/api-inference/tasks/image-classification.md

-1
Original file line numberDiff line numberDiff line change
@@ -25,7 +25,6 @@ For more details about the `image-classification` task, check out its [dedicated
2525
### Recommended models
2626

2727
- [google/vit-base-patch16-224](https://huggingface.co/google/vit-base-patch16-224): A strong image classification model.
28-
- [facebook/deit-base-distilled-patch16-224](https://huggingface.co/facebook/deit-base-distilled-patch16-224): A robust image classification model.
2928

3029
Explore all available models and find the one that suits you best [here](https://huggingface.co/models?inference=warm&pipeline_tag=image-classification&sort=trending).
3130

0 commit comments

Comments
 (0)