File size: 3,430 Bytes
8de92d1
ff1e468
8de92d1
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
ff1e468
8de92d1
 
 
ff1e468
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
70a8dcb
ff1e468
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1a43738
ff1e468
 
 
 
1a43738
ff1e468
 
 
 
 
 
 
 
 
 
 
 
 
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
import { InferenceClient } from "@huggingface/inference";
import { useState } from "react";

export default async function inference({
  prompt,
  model = "Qwen/Qwen3-235B-A22B",
  apiKey,
  maxTokens = 512
}: {
  prompt: string,
  model?: string,
  apiKey?: string,
  maxTokens?: number
}) {
  if (!apiKey) {
    const token = window.localStorage.getItem("huggingface_access_token");
    if (!token) {
      throw new Error("You must be signed in to use the inference API!");
    }
    apiKey = token;
  }

  console.log("Inference", prompt, model, apiKey);
  const client = new InferenceClient(apiKey);

  const chatCompletion = await client.chatCompletion({
    provider: "fireworks-ai",
    model: model,
    messages: [
      {
        role: "user",
        content: prompt,
      },
    ],
    max_tokens: maxTokens,
  });
  

  console.log("Inference response", chatCompletion.choices[0].message);
  return chatCompletion.choices[0].message;
}

export function useInferenceOld({ apiKey }) {
  const [isLoading, setIsLoading] = useState(false);
  const [partialText, setPartialText] = useState("");
  const [inferenceResult, setInferenceResult] = useState("");
  const [error, setError] = useState<string | null>(null);
  const inferenceInternal = async ({
    prompt,
    model,
    maxTokens,
  }: {
    prompt: string;
    model: string;
    maxTokens: number;
  }) => {
    setIsLoading(true);
    setPartialText("boop boop partial text");

    try {
      const result = await inference({
        prompt,
        model,
        apiKey,
        maxTokens,
      });

      setInferenceResult(result.content);
      setIsLoading(false);

      return result.content;
    } catch (error) {
      console.error("Error in inference", error);
      setError(error.message);
      setIsLoading(false);
      return null;
    }
  };

  const status = isLoading ? "thinking" : error ? "error" : "done";

  return {
    status,
    partialText,
    inferenceResult,
    error,
    inference: inferenceInternal,
  };
}


export function useInference({ apiKey }) {
  const [isLoading, setIsLoading] = useState(false);
  const [partialText, setPartialText] = useState("");
  const [inferenceResult, setInferenceResult] = useState("");
  const [error, setError] = useState<string | null>(null);
  const inferenceInternal = async ({
    prompt,
    model,
    maxTokens,
  }: {
    prompt: string;
    model: string;
    maxTokens: number;
  }) => {
    setIsLoading(true);
    setPartialText("");

    const client = new InferenceClient(apiKey);

    try {
      const stream = client.chatCompletionStream({
        provider: "nebius",
        model,
        maxTokens,
        messages: [
          {
            role: "user",
            content: prompt,
          },
        ],
      });

      let result = "";

      for await (const chunk of stream) {
        result += chunk.choices[0].delta.content;
        setPartialText(result);
      }

      setIsLoading(false);

      setInferenceResult(result);

      return {status: "success", result};
    } catch (error) {
      console.error("Error in inference", error);
      setError(error.message);
      setIsLoading(false);
      return {status: "error", result: error.message};
    }
  };

  const status = isLoading ? "thinking" : error ? "error" : "done";

  return {
    status,
    partialText,
    inferenceResult,
    error,
    inference: inferenceInternal,
  };
}