add conversate v2 and v3

#15
by ereniko - opened
Files changed (1) hide show
  1. app.py +11 -0
app.py CHANGED
@@ -74,6 +74,8 @@ BASE_MODEL_IDS: List[str] = [
74
  "DedeProGames/Kiyo-230M-Preview",
75
  "DedeProGames/Kiyo-135M",
76
  "DedeProGames/Kiyo-65M",
 
 
77
  ]
78
 
79
  BASE_MODEL_DISPLAY: Dict[str, str] = {
@@ -93,6 +95,8 @@ BASE_MODEL_DISPLAY: Dict[str, str] = {
93
  "DedeProGames/Kiyo-230M-Preview": "Kiyo-230M-Preview",
94
  "DedeProGames/Kiyo-135M": "Kiyo-135M",
95
  "DedeProGames/Kiyo-65M": "Kiyo-65M",
 
 
96
  }
97
 
98
  MODEL_PARAMS: Dict[str, float] = {
@@ -124,6 +128,8 @@ MODEL_PARAMS: Dict[str, float] = {
124
  "DedeProGames/Kiyo-135M": 134.5e6,
125
  "DedeProGames/Kiyo-65M": 65.0e6,
126
  "wayneworkman2012/peacebell-v1-148M": 148.55e6,
 
 
127
  }
128
 
129
  FALLBACK_IDS: Dict[str, str] = {}
@@ -223,6 +229,8 @@ GEN_DEFAULTS: Dict[str, dict] = {
223
  "DedeProGames/Kiyo-65M": {"max_new_tokens": 64, "temperature": 0.8, "top_p": 0.95, "repetition_penalty": 1.1, "do_sample": True},
224
  # Peacebell: greedy (temperature 0), no repetition penalty, no n-gram block; decoded by its own ChatML path (generate_peacebell)
225
  "wayneworkman2012/peacebell-v1-148M": {"max_new_tokens": 64, "temperature": 0.0, "do_sample": False, "repetition_penalty": 1.0},
 
 
226
  }
227
 
228
  MODEL_CONTEXT: Dict[str, int] = {
@@ -254,6 +262,9 @@ MODEL_CONTEXT: Dict[str, int] = {
254
  "DedeProGames/Kiyo-135M": 8192,
255
  "DedeProGames/Kiyo-65M": 2048,
256
  "wayneworkman2012/peacebell-v1-148M": 16384,
 
 
 
257
  }
258
 
259
 
 
74
  "DedeProGames/Kiyo-230M-Preview",
75
  "DedeProGames/Kiyo-135M",
76
  "DedeProGames/Kiyo-65M",
77
+ "IvmeLabs/Ivme-Conversate-v2-Base",
78
+ "IvmeLabs/Ivme-Conversate-v3-Base",
79
  ]
80
 
81
  BASE_MODEL_DISPLAY: Dict[str, str] = {
 
95
  "DedeProGames/Kiyo-230M-Preview": "Kiyo-230M-Preview",
96
  "DedeProGames/Kiyo-135M": "Kiyo-135M",
97
  "DedeProGames/Kiyo-65M": "Kiyo-65M",
98
+ "IvmeLabs/Ivme-Conversate-v2-Base": "Ivme-Conversate-v2-Base",
99
+ "IvmeLabs/Ivme-Conversate-v3-Base": "Ivme-Conversate-v3-Base",
100
  }
101
 
102
  MODEL_PARAMS: Dict[str, float] = {
 
128
  "DedeProGames/Kiyo-135M": 134.5e6,
129
  "DedeProGames/Kiyo-65M": 65.0e6,
130
  "wayneworkman2012/peacebell-v1-148M": 148.55e6,
131
+ "IvmeLabs/Ivme-Conversate-v2-Base": 23.8e6,
132
+ "IvmeLabs/Ivme-Conversate-v3-Base": 24.8e6,
133
  }
134
 
135
  FALLBACK_IDS: Dict[str, str] = {}
 
229
  "DedeProGames/Kiyo-65M": {"max_new_tokens": 64, "temperature": 0.8, "top_p": 0.95, "repetition_penalty": 1.1, "do_sample": True},
230
  # Peacebell: greedy (temperature 0), no repetition penalty, no n-gram block; decoded by its own ChatML path (generate_peacebell)
231
  "wayneworkman2012/peacebell-v1-148M": {"max_new_tokens": 64, "temperature": 0.0, "do_sample": False, "repetition_penalty": 1.0},
232
+ "IvmeLabs/Ivme-Conversate-v2-Base": {"max_new_tokens": 64, "temperature": 0.8, "top_p": 0.95, "repetition_penalty": 1.1, "do_sample": True},
233
+ "IvmeLabs/Ivme-Conversate-v3-Base": {"max_new_tokens": 64, "temperature": 0.8, "top_p": 0.95, "repetition_penalty": 1.1, "do_sample": True},
234
  }
235
 
236
  MODEL_CONTEXT: Dict[str, int] = {
 
262
  "DedeProGames/Kiyo-135M": 8192,
263
  "DedeProGames/Kiyo-65M": 2048,
264
  "wayneworkman2012/peacebell-v1-148M": 16384,
265
+ "IvmeLabs/Ivme-Conversate-v2-Base": 1024,
266
+ "IvmeLabs/Ivme-Conversate-v3-Base": 1024,
267
+
268
  }
269
 
270