Update app.py
Browse files
app.py
CHANGED
@@ -22,20 +22,6 @@ dante = Llama(
|
|
22 |
n_ctx=4086,
|
23 |
)
|
24 |
|
25 |
-
karg = {
|
26 |
-
'input_prompt': input_prompt,
|
27 |
-
'temperature': 0.15,
|
28 |
-
'top_p': 0.1,
|
29 |
-
'top_k': 40,
|
30 |
-
'repeat_penalty': 1.1,
|
31 |
-
'max_tokens': 1024,
|
32 |
-
'stop': [
|
33 |
-
"[|Umano|]",
|
34 |
-
"[|Assistente|]",
|
35 |
-
],
|
36 |
-
'stream': True
|
37 |
-
}
|
38 |
-
|
39 |
history = []
|
40 |
|
41 |
def generate_text(message, history):
|
@@ -49,6 +35,19 @@ def generate_text(message, history):
|
|
49 |
|
50 |
print(input_prompt)
|
51 |
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
52 |
output = saiga(**karg)
|
53 |
|
54 |
for out in output:
|
@@ -68,6 +67,20 @@ def generate_text_Dante(message, history):
|
|
68 |
input_prompt += "[|Umano|] " + message + "\n[|Assistente|]"
|
69 |
|
70 |
print(input_prompt)
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
71 |
|
72 |
output = dante(**karg)
|
73 |
|
|
|
22 |
n_ctx=4086,
|
23 |
)
|
24 |
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
25 |
history = []
|
26 |
|
27 |
def generate_text(message, history):
|
|
|
35 |
|
36 |
print(input_prompt)
|
37 |
|
38 |
+
karg = {
|
39 |
+
'input_prompt': input_prompt,
|
40 |
+
'temperature': 0.15,
|
41 |
+
'top_p': 0.1,
|
42 |
+
'top_k': 40,
|
43 |
+
'repeat_penalty': 1.1,
|
44 |
+
'max_tokens': 1024,
|
45 |
+
'stop': [
|
46 |
+
"[|Umano|]",
|
47 |
+
"[|Assistente|]",
|
48 |
+
],
|
49 |
+
'stream': True
|
50 |
+
}
|
51 |
output = saiga(**karg)
|
52 |
|
53 |
for out in output:
|
|
|
67 |
input_prompt += "[|Umano|] " + message + "\n[|Assistente|]"
|
68 |
|
69 |
print(input_prompt)
|
70 |
+
|
71 |
+
karg = {
|
72 |
+
'input_prompt': input_prompt,
|
73 |
+
'temperature': 0.15,
|
74 |
+
'top_p': 0.1,
|
75 |
+
'top_k': 40,
|
76 |
+
'repeat_penalty': 1.1,
|
77 |
+
'max_tokens': 1024,
|
78 |
+
'stop': [
|
79 |
+
"[|Umano|]",
|
80 |
+
"[|Assistente|]",
|
81 |
+
],
|
82 |
+
'stream': True
|
83 |
+
}
|
84 |
|
85 |
output = dante(**karg)
|
86 |
|