Hugging Face's logo

Spaces:
emekaboris
/
try-this-model
Running

App Files Community
try-this-model
File size: 4,316 Bytes 
e0ccd06
 
 
 
 
a1c24d9
 
e0ccd06
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
77ee232
 
 
 
 
e0ccd06
 
 
 
 
 
 
 
 
 
 
 
 
2018dd8
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
e0ccd06
 
 
2018dd8
 
 
 
 
5ae724e
e0ccd06
 
 
 
 
 
 
a1c24d9
e0ccd06
a1c24d9
 
e0ccd06
 
 
 
 
 
 
a1c24d9
 
 
 
e0ccd06
 
34e11d5
 
 
 
 
 
 
 
 
 
 
 
 
e0ccd06
 
 
 
 
e00ab88
 
 
e0ccd06
 
 
 
 
e00ab88
e0ccd06
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
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
 
from openai import OpenAI
import gradio as gr
import os
import json
import functools
import random
import datetime

api_key = os.environ.get('FEATHERLESS_API_KEY')
client = OpenAI(
    base_url="https://api.featherless.ai/v1",
    api_key=api_key
)

def respond(message, history, model):
    history_openai_format = []
    for human, assistant in history:
        history_openai_format.append({"role": "user", "content": human })
        history_openai_format.append({"role": "assistant", "content":assistant})
    history_openai_format.append({"role": "user", "content": message})
  
    response = client.chat.completions.create(
        model=model,
        messages= history_openai_format,
        temperature=1.0,
        stream=True,
        max_tokens=2000,
        extra_headers={
            'HTTP-Referer': 'https://huggingface.co/spaces/featherless-ai/try-this-model',
            'X-Title': "HF's missing inference widget"
        }
    )

    partial_message = ""
    for chunk in response:
        if chunk.choices[0].delta.content is not None:
              partial_message = partial_message + chunk.choices[0].delta.content
              yield partial_message

logo = open('./logo.svg').read()

with open('./model-cache.json', 'r') as f_model_cache:
    model_cache = json.load(f_model_cache)


model_class_filter = {
    "mistral-v02-7b-std-lc": True,
    "llama3-8b-8k": True,
    "llama2-solar-10b7-4k": True,
    "mistral-nemo-12b-lc": True,
    "llama2-13b-4k": True,
    "llama3-15b-8k": True,

    "qwen2-32b-lc":False,
    "llama3-70b-8k":False,
    "qwen2-72b-lc":False,
    "mixtral-8x22b-lc":False,
    "llama3-405b-lc":False,
}

def build_model_choices():
    all_choices = []
    for model_class in model_cache:
        if model_class not in model_class_filter:
            print(f"Warning: new model class {model_class}. Treating as blacklisted")
            continue

        if not model_class_filter[model_class]:
            continue
        all_choices += [ (f"{model_id} ({model_class})", model_id) for model_id in model_cache[model_class] ]

    return all_choices

model_choices = build_model_choices()

def initial_model(referer=None):

    if referer == 'http://127.0.0.1:7860/':
        return 'Sao10K/Venomia-1.1-m7'

    if referer and referer.startswith("https://huggingface.co/"):
        possible_model = referer[23:]
        full_model_list = functools.reduce(lambda x,y: x+y, model_cache.values(), [])
        model_is_supported = possible_model in full_model_list
        if model_is_supported:
            return possible_model

    # let's use a random but different model each day.
    key=os.environ.get('RANDOM_SEED', 'kcOtfNHA+e')
    o = random.Random(f"{key}-{datetime.date.today().strftime('%Y-%m-%d')}")
    return o.choice(model_choices)[1]

title_text="HuggingFace's missing inference widget"
css = """
.logo-mark { fill: #ffe184; }

/* from https://github.com/gradio-app/gradio/issues/4001
 * necessary as putting ChatInterface in gr.Blocks changes behaviour
 */

.contain { display: flex; flex-direction: column; }
.gradio-container { height: 100vh !important; }
#component-0 { height: 100%; }
#chatbot { flex-grow: 1; overflow: auto;}
"""
with gr.Blocks(title_text, css=css) as demo:
    gr.HTML("""
        <h1 align="center">HuggingFace's missing inference widget</h1>
        <p align="center">
            Test any <=15B LLM from the hub.
        </p>
        <h2 align="center">
            Please select your model from the list 👇 as HF spaces can't see the refering model card.
        </h2>
    """)

    # hidden_state = gr.State(value=initial_model)

    model_selector = gr.Dropdown(
        label="Select your Model",
        choices=build_model_choices(),
        value=initial_model
        # value=hidden_state
    )

    gr.ChatInterface(
        respond,
        additional_inputs=[model_selector],
        head=""",
        <script>console.log("Hello from gradio!")</script>
        """,
    )
    gr.HTML(f"""
        <p align="center">
            Inference by <a href="https://featherless.ai">{logo}</a>
        </p>
    """)
    def update_initial_model_choice(request: gr.Request):
        return initial_model(request.headers.get('referer'))

    demo.load(update_initial_model_choice, outputs=model_selector)

demo.launch()