mirza176528470100 commited on
Commit
c61c4a7
·
verified ·
1 Parent(s): e40306e

Upload 4 files

Browse files
Files changed (4) hide show
  1. app.py +89 -70
  2. model.pth +3 -0
  3. names.txt +0 -0
  4. requirements.txt +2 -0
app.py CHANGED
@@ -1,70 +1,89 @@
1
- import gradio as gr
2
- from huggingface_hub import InferenceClient
3
-
4
-
5
- def respond(
6
- message,
7
- history: list[dict[str, str]],
8
- system_message,
9
- max_tokens,
10
- temperature,
11
- top_p,
12
- hf_token: gr.OAuthToken,
13
- ):
14
- """
15
- For more information on `huggingface_hub` Inference API support, please check the docs: https://huggingface.co/docs/huggingface_hub/v0.22.2/en/guides/inference
16
- """
17
- client = InferenceClient(token=hf_token.token, model="openai/gpt-oss-20b")
18
-
19
- messages = [{"role": "system", "content": system_message}]
20
-
21
- messages.extend(history)
22
-
23
- messages.append({"role": "user", "content": message})
24
-
25
- response = ""
26
-
27
- for message in client.chat_completion(
28
- messages,
29
- max_tokens=max_tokens,
30
- stream=True,
31
- temperature=temperature,
32
- top_p=top_p,
33
- ):
34
- choices = message.choices
35
- token = ""
36
- if len(choices) and choices[0].delta.content:
37
- token = choices[0].delta.content
38
-
39
- response += token
40
- yield response
41
-
42
-
43
- """
44
- For information on how to customize the ChatInterface, peruse the gradio docs: https://www.gradio.app/docs/chatinterface
45
- """
46
- chatbot = gr.ChatInterface(
47
- respond,
48
- type="messages",
49
- additional_inputs=[
50
- gr.Textbox(value="You are a friendly Chatbot.", label="System message"),
51
- gr.Slider(minimum=1, maximum=2048, value=512, step=1, label="Max new tokens"),
52
- gr.Slider(minimum=0.1, maximum=4.0, value=0.7, step=0.1, label="Temperature"),
53
- gr.Slider(
54
- minimum=0.1,
55
- maximum=1.0,
56
- value=0.95,
57
- step=0.05,
58
- label="Top-p (nucleus sampling)",
59
- ),
60
- ],
61
- )
62
-
63
- with gr.Blocks() as demo:
64
- with gr.Sidebar():
65
- gr.LoginButton()
66
- chatbot.render()
67
-
68
-
69
- if __name__ == "__main__":
70
- demo.launch()
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ import torch
2
+ import torch.nn.functional as F
3
+ import gradio as gr
4
+ import os
5
+
6
+ with open('names.txt', 'r') as f:
7
+ words = f.read().splitlines()
8
+
9
+ chars = sorted(list(set(''.join(words))))
10
+ stoi = {s:i+1 for i,s in enumerate(chars)}
11
+ stoi['.'] = 0
12
+ itos = {i:s for s,i in stoi.items()}
13
+ vocab_size = len(itos)
14
+ block_size = 3
15
+
16
+ class ProductionMLP(torch.nn.Module):
17
+ def __init__(self, vocab_size, n_embd, block_size, n_hidden):
18
+ super().__init__()
19
+ self.embedding = torch.nn.Embedding(vocab_size, n_embd)
20
+ self.layers = torch.nn.Sequential(
21
+ torch.nn.Linear(n_embd * block_size, n_hidden, bias=False),
22
+ torch.nn.BatchNorm1d(n_hidden),
23
+ torch.nn.Tanh(),
24
+ torch.nn.Linear(n_hidden, vocab_size)
25
+ )
26
+
27
+ def forward(self, x):
28
+ emb = self.embedding(x)
29
+ emb = emb.view(emb.shape[0], -1)
30
+ logits = self.layers(emb)
31
+ return logits
32
+
33
+
34
+ n_embd = 10
35
+ n_hidden = 200
36
+
37
+ model = ProductionMLP(vocab_size, n_embd, block_size, n_hidden)
38
+
39
+
40
+ model.load_state_dict(torch.load("model.pth", map_location=torch.device('cpu')))
41
+ model.eval()
42
+
43
+
44
+ def generate_name(start_letter):
45
+
46
+ if not start_letter or start_letter.strip() == "":
47
+ start_letter = '.'
48
+
49
+ start_letter = start_letter.lower()
50
+
51
+ if start_letter not in stoi:
52
+ return f"Error: '{start_letter}' is not in the vocabulary. Please use a-z."
53
+
54
+
55
+ context = [0] * block_size
56
+ output_name = []
57
+
58
+
59
+ if start_letter != '.':
60
+ output_name.append(start_letter)
61
+ idx = stoi[start_letter]
62
+ context = context[1:] + [idx]
63
+
64
+
65
+ while True:
66
+ x = torch.tensor([context])
67
+ with torch.no_grad():
68
+ logits = model(x)
69
+ probs = F.softmax(logits, dim=1)
70
+ ix = torch.multinomial(probs, num_samples=1).item()
71
+
72
+ if ix == 0:
73
+ break
74
+
75
+ context = context[1:] + [ix]
76
+ output_name.append(itos[ix])
77
+
78
+ return "".join(output_name).capitalize()
79
+
80
+
81
+ iface = gr.Interface(
82
+ fn=generate_name,
83
+ inputs=gr.Textbox(label="Start with a letter? (Leave blank for random)", placeholder="e.g. m"),
84
+ outputs="text",
85
+ title="Neural Name Generator",
86
+ description="This model was trained on thousands of names. Give it a starting letter, or leave it blank to see what it invents!"
87
+ )
88
+
89
+ iface.launch()
model.pth ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:45410bf48c383ae924b7509cada036812108dc94edc13d26c139172b1d051190
3
+ size 53547
names.txt ADDED
The diff for this file is too large to render. See raw diff
 
requirements.txt ADDED
@@ -0,0 +1,2 @@
 
 
 
1
+ torch
2
+ gradio