Run 5. Outer Step 70. Inner Step 1.
Browse files- config.json +82 -82
- inner_optimizer.pt +1 -1
- model.safetensors +1 -1
config.json
CHANGED
|
@@ -8,28 +8,28 @@
|
|
| 8 |
"101": "NON_PARTICIPATING",
|
| 9 |
"102": "NON_PARTICIPATING",
|
| 10 |
"103": "SUCCESS",
|
| 11 |
-
"104": "
|
| 12 |
"105": "NON_PARTICIPATING",
|
| 13 |
-
"106": "
|
| 14 |
"107": "NON_PARTICIPATING",
|
| 15 |
"108": "NON_PARTICIPATING",
|
| 16 |
-
"109": "
|
| 17 |
-
"11": "
|
| 18 |
"110": "SUCCESS",
|
| 19 |
-
"111": "
|
| 20 |
"112": "NON_PARTICIPATING",
|
| 21 |
"113": "NON_PARTICIPATING",
|
| 22 |
-
"114": "
|
| 23 |
"115": "NOT_ALIVE",
|
| 24 |
"116": "NON_PARTICIPATING",
|
| 25 |
"117": "NON_PARTICIPATING",
|
| 26 |
-
"118": "
|
| 27 |
"119": "NOT_ALIVE",
|
| 28 |
-
"12": "
|
| 29 |
"120": "NON_PARTICIPATING",
|
| 30 |
-
"121": "
|
| 31 |
"122": "SUCCESS",
|
| 32 |
-
"123": "
|
| 33 |
"124": "NON_PARTICIPATING",
|
| 34 |
"125": "NON_PARTICIPATING",
|
| 35 |
"126": "NON_PARTICIPATING",
|
|
@@ -37,77 +37,77 @@
|
|
| 37 |
"128": "NOT_ALIVE",
|
| 38 |
"129": "NON_PARTICIPATING",
|
| 39 |
"13": "NON_PARTICIPATING",
|
| 40 |
-
"130": "
|
| 41 |
"131": "NOT_ALIVE",
|
| 42 |
"132": "NOT_ALIVE",
|
| 43 |
"133": "SUCCESS",
|
| 44 |
"134": "NOT_ALIVE",
|
| 45 |
-
"135": "
|
| 46 |
"136": "SUCCESS",
|
| 47 |
-
"137": "
|
| 48 |
"138": "NON_PARTICIPATING",
|
| 49 |
-
"139": "
|
| 50 |
-
"14": "
|
| 51 |
-
"140": "
|
| 52 |
-
"141": "
|
| 53 |
-
"142": "
|
| 54 |
"143": "NON_PARTICIPATING",
|
| 55 |
"144": "NON_PARTICIPATING",
|
| 56 |
-
"145": "
|
| 57 |
"146": "NON_PARTICIPATING",
|
| 58 |
"147": "NON_PARTICIPATING",
|
| 59 |
"148": "SUCCESS",
|
| 60 |
-
"149": "
|
| 61 |
"15": "NON_PARTICIPATING",
|
| 62 |
-
"150": "
|
| 63 |
"151": "NOT_ALIVE",
|
| 64 |
"152": "NON_PARTICIPATING",
|
| 65 |
-
"153": "
|
| 66 |
"154": "NON_PARTICIPATING",
|
| 67 |
"155": "NON_PARTICIPATING",
|
| 68 |
"156": "SUCCESS",
|
| 69 |
"157": "NON_PARTICIPATING",
|
| 70 |
-
"158": "
|
| 71 |
"159": "NOT_ALIVE",
|
| 72 |
"16": "NON_PARTICIPATING",
|
| 73 |
-
"160": "
|
| 74 |
"161": "NON_PARTICIPATING",
|
| 75 |
"162": "NON_PARTICIPATING",
|
| 76 |
-
"163": "
|
| 77 |
"164": "SUCCESS",
|
| 78 |
"165": "SUCCESS",
|
| 79 |
"166": "NON_PARTICIPATING",
|
| 80 |
-
"167": "
|
| 81 |
-
"168": "
|
| 82 |
"169": "NON_PARTICIPATING",
|
| 83 |
-
"17": "
|
| 84 |
"170": "NON_PARTICIPATING",
|
| 85 |
-
"171": "
|
| 86 |
"172": "NON_PARTICIPATING",
|
| 87 |
-
"173": "
|
| 88 |
"174": "NON_PARTICIPATING",
|
| 89 |
-
"175": "
|
| 90 |
"176": "NOT_ALIVE",
|
| 91 |
"177": "SUCCESS",
|
| 92 |
"178": "NOT_ALIVE",
|
| 93 |
-
"179": "
|
| 94 |
-
"18": "
|
| 95 |
-
"180": "
|
| 96 |
"181": "SUCCESS",
|
| 97 |
-
"182": "
|
| 98 |
"183": "NON_PARTICIPATING",
|
| 99 |
-
"184": "
|
| 100 |
"185": "NON_PARTICIPATING",
|
| 101 |
-
"186": "
|
| 102 |
"187": "NON_PARTICIPATING",
|
| 103 |
-
"188": "
|
| 104 |
"189": "NOT_ALIVE",
|
| 105 |
"19": "NOT_ALIVE",
|
| 106 |
"190": "NOT_ALIVE",
|
| 107 |
"191": "NOT_ALIVE",
|
| 108 |
"192": "NON_PARTICIPATING",
|
| 109 |
"193": "NON_PARTICIPATING",
|
| 110 |
-
"194": "
|
| 111 |
"195": "NOT_ALIVE",
|
| 112 |
"196": "NOT_ALIVE",
|
| 113 |
"197": "NON_PARTICIPATING",
|
|
@@ -119,9 +119,9 @@
|
|
| 119 |
"201": "NON_PARTICIPATING",
|
| 120 |
"202": "NON_PARTICIPATING",
|
| 121 |
"203": "NON_PARTICIPATING",
|
| 122 |
-
"204": "
|
| 123 |
"205": "NON_PARTICIPATING",
|
| 124 |
-
"206": "
|
| 125 |
"207": "NOT_ALIVE",
|
| 126 |
"208": "NOT_ALIVE",
|
| 127 |
"209": "NON_PARTICIPATING",
|
|
@@ -129,46 +129,46 @@
|
|
| 129 |
"210": "SUCCESS",
|
| 130 |
"211": "NON_PARTICIPATING",
|
| 131 |
"212": "NOT_ALIVE",
|
| 132 |
-
"213": "
|
| 133 |
"214": "NON_PARTICIPATING",
|
| 134 |
-
"215": "
|
| 135 |
-
"216": "
|
| 136 |
-
"217": "
|
| 137 |
"218": "NON_PARTICIPATING",
|
| 138 |
"219": "NON_PARTICIPATING",
|
| 139 |
"22": "NON_PARTICIPATING",
|
| 140 |
"220": "NON_PARTICIPATING",
|
| 141 |
"221": "SUCCESS",
|
| 142 |
-
"222": "
|
| 143 |
-
"223": "
|
| 144 |
"224": "NOT_ALIVE",
|
| 145 |
"225": "SUCCESS",
|
| 146 |
-
"226": "
|
| 147 |
"227": "NON_PARTICIPATING",
|
| 148 |
"228": "NON_PARTICIPATING",
|
| 149 |
-
"229": "
|
| 150 |
"23": "NON_PARTICIPATING",
|
| 151 |
-
"230": "
|
| 152 |
"231": "NON_PARTICIPATING",
|
| 153 |
"232": "NON_PARTICIPATING",
|
| 154 |
-
"233": "
|
| 155 |
"234": "NOT_ALIVE",
|
| 156 |
-
"235": "
|
| 157 |
"236": "NON_PARTICIPATING",
|
| 158 |
-
"237": "
|
| 159 |
"238": "SUCCESS",
|
| 160 |
"239": "NON_PARTICIPATING",
|
| 161 |
"24": "SUCCESS",
|
| 162 |
"240": "NON_PARTICIPATING",
|
| 163 |
-
"241": "
|
| 164 |
"242": "SUCCESS",
|
| 165 |
"243": "NON_PARTICIPATING",
|
| 166 |
"244": "NON_PARTICIPATING",
|
| 167 |
-
"245": "
|
| 168 |
"246": "NON_PARTICIPATING",
|
| 169 |
"247": "NON_PARTICIPATING",
|
| 170 |
"248": "NON_PARTICIPATING",
|
| 171 |
-
"249": "
|
| 172 |
"25": "SUCCESS",
|
| 173 |
"250": "NON_PARTICIPATING",
|
| 174 |
"251": "NOT_ALIVE",
|
|
@@ -178,19 +178,19 @@
|
|
| 178 |
"255": "NON_PARTICIPATING",
|
| 179 |
"26": "NON_PARTICIPATING",
|
| 180 |
"27": "SUCCESS",
|
| 181 |
-
"28": "
|
| 182 |
-
"29": "
|
| 183 |
"3": "NON_PARTICIPATING",
|
| 184 |
"30": "SUCCESS",
|
| 185 |
"31": "NON_PARTICIPATING",
|
| 186 |
"32": "SUCCESS",
|
| 187 |
"33": "NOT_ALIVE",
|
| 188 |
-
"34": "
|
| 189 |
-
"35": "
|
| 190 |
"36": "NON_PARTICIPATING",
|
| 191 |
"37": "NON_PARTICIPATING",
|
| 192 |
"38": "SUCCESS",
|
| 193 |
-
"39": "
|
| 194 |
"4": "NON_PARTICIPATING",
|
| 195 |
"40": "SUCCESS",
|
| 196 |
"41": "SUCCESS",
|
|
@@ -209,53 +209,53 @@
|
|
| 209 |
"53": "NOT_ALIVE",
|
| 210 |
"54": "NON_PARTICIPATING",
|
| 211 |
"55": "NON_PARTICIPATING",
|
| 212 |
-
"56": "
|
| 213 |
-
"57": "
|
| 214 |
-
"58": "
|
| 215 |
"59": "NON_PARTICIPATING",
|
| 216 |
-
"6": "
|
| 217 |
"60": "NON_PARTICIPATING",
|
| 218 |
"61": "NOT_ALIVE",
|
| 219 |
-
"62": "
|
| 220 |
"63": "NON_PARTICIPATING",
|
| 221 |
"64": "NON_PARTICIPATING",
|
| 222 |
"65": "NOT_ALIVE",
|
| 223 |
-
"66": "
|
| 224 |
-
"67": "
|
| 225 |
"68": "NON_PARTICIPATING",
|
| 226 |
"69": "NON_PARTICIPATING",
|
| 227 |
"7": "NON_PARTICIPATING",
|
| 228 |
-
"70": "
|
| 229 |
-
"71": "
|
| 230 |
-
"72": "
|
| 231 |
"73": "NON_PARTICIPATING",
|
| 232 |
"74": "NON_PARTICIPATING",
|
| 233 |
"75": "SUCCESS",
|
| 234 |
"76": "NON_PARTICIPATING",
|
| 235 |
"77": "NOT_ALIVE",
|
| 236 |
-
"78": "
|
| 237 |
"79": "NON_PARTICIPATING",
|
| 238 |
"8": "NON_PARTICIPATING",
|
| 239 |
"80": "NON_PARTICIPATING",
|
| 240 |
"81": "NON_PARTICIPATING",
|
| 241 |
"82": "NOT_ALIVE",
|
| 242 |
"83": "NON_PARTICIPATING",
|
| 243 |
-
"84": "
|
| 244 |
"85": "NOT_ALIVE",
|
| 245 |
-
"86": "
|
| 246 |
"87": "NON_PARTICIPATING",
|
| 247 |
-
"88": "
|
| 248 |
"89": "NON_PARTICIPATING",
|
| 249 |
"9": "NON_PARTICIPATING",
|
| 250 |
"90": "SUCCESS",
|
| 251 |
-
"91": "
|
| 252 |
-
"92": "
|
| 253 |
"93": "NON_PARTICIPATING",
|
| 254 |
"94": "NON_PARTICIPATING",
|
| 255 |
"95": "NON_PARTICIPATING",
|
| 256 |
"96": "NON_PARTICIPATING",
|
| 257 |
"97": "NON_PARTICIPATING",
|
| 258 |
-
"98": "
|
| 259 |
"99": "NON_PARTICIPATING"
|
| 260 |
},
|
| 261 |
"architectures": [
|
|
@@ -264,17 +264,17 @@
|
|
| 264 |
"attention_bias": false,
|
| 265 |
"attention_dropout": 0.0,
|
| 266 |
"block_list": [
|
| 267 |
-
|
| 268 |
-
|
| 269 |
],
|
| 270 |
"bos_token_id": 1,
|
| 271 |
"eos_token_id": 2,
|
| 272 |
"hidden_act": "silu",
|
| 273 |
"hidden_size": 2048,
|
| 274 |
"initializer_range": 0.02,
|
| 275 |
-
"inner_step":
|
| 276 |
"intermediate_size": 5632,
|
| 277 |
-
"last_allreduce_block":
|
| 278 |
"max_position_embeddings": 2048,
|
| 279 |
"mlp_bias": false,
|
| 280 |
"model_type": "llama",
|
|
|
|
| 8 |
"101": "NON_PARTICIPATING",
|
| 9 |
"102": "NON_PARTICIPATING",
|
| 10 |
"103": "SUCCESS",
|
| 11 |
+
"104": "NOT_ALIVE",
|
| 12 |
"105": "NON_PARTICIPATING",
|
| 13 |
+
"106": "NOT_ALIVE",
|
| 14 |
"107": "NON_PARTICIPATING",
|
| 15 |
"108": "NON_PARTICIPATING",
|
| 16 |
+
"109": "NON_PARTICIPATING",
|
| 17 |
+
"11": "NON_PARTICIPATING",
|
| 18 |
"110": "SUCCESS",
|
| 19 |
+
"111": "NOT_ALIVE",
|
| 20 |
"112": "NON_PARTICIPATING",
|
| 21 |
"113": "NON_PARTICIPATING",
|
| 22 |
+
"114": "NON_PARTICIPATING",
|
| 23 |
"115": "NOT_ALIVE",
|
| 24 |
"116": "NON_PARTICIPATING",
|
| 25 |
"117": "NON_PARTICIPATING",
|
| 26 |
+
"118": "NOT_ALIVE",
|
| 27 |
"119": "NOT_ALIVE",
|
| 28 |
+
"12": "NOT_ALIVE",
|
| 29 |
"120": "NON_PARTICIPATING",
|
| 30 |
+
"121": "NON_PARTICIPATING",
|
| 31 |
"122": "SUCCESS",
|
| 32 |
+
"123": "NON_PARTICIPATING",
|
| 33 |
"124": "NON_PARTICIPATING",
|
| 34 |
"125": "NON_PARTICIPATING",
|
| 35 |
"126": "NON_PARTICIPATING",
|
|
|
|
| 37 |
"128": "NOT_ALIVE",
|
| 38 |
"129": "NON_PARTICIPATING",
|
| 39 |
"13": "NON_PARTICIPATING",
|
| 40 |
+
"130": "NOT_ALIVE",
|
| 41 |
"131": "NOT_ALIVE",
|
| 42 |
"132": "NOT_ALIVE",
|
| 43 |
"133": "SUCCESS",
|
| 44 |
"134": "NOT_ALIVE",
|
| 45 |
+
"135": "NON_PARTICIPATING",
|
| 46 |
"136": "SUCCESS",
|
| 47 |
+
"137": "NOT_ALIVE",
|
| 48 |
"138": "NON_PARTICIPATING",
|
| 49 |
+
"139": "NOT_ALIVE",
|
| 50 |
+
"14": "NOT_ALIVE",
|
| 51 |
+
"140": "NOT_ALIVE",
|
| 52 |
+
"141": "NON_PARTICIPATING",
|
| 53 |
+
"142": "SUCCESS",
|
| 54 |
"143": "NON_PARTICIPATING",
|
| 55 |
"144": "NON_PARTICIPATING",
|
| 56 |
+
"145": "NOT_ALIVE",
|
| 57 |
"146": "NON_PARTICIPATING",
|
| 58 |
"147": "NON_PARTICIPATING",
|
| 59 |
"148": "SUCCESS",
|
| 60 |
+
"149": "NOT_ALIVE",
|
| 61 |
"15": "NON_PARTICIPATING",
|
| 62 |
+
"150": "NOT_ALIVE",
|
| 63 |
"151": "NOT_ALIVE",
|
| 64 |
"152": "NON_PARTICIPATING",
|
| 65 |
+
"153": "SUCCESS",
|
| 66 |
"154": "NON_PARTICIPATING",
|
| 67 |
"155": "NON_PARTICIPATING",
|
| 68 |
"156": "SUCCESS",
|
| 69 |
"157": "NON_PARTICIPATING",
|
| 70 |
+
"158": "NOT_ALIVE",
|
| 71 |
"159": "NOT_ALIVE",
|
| 72 |
"16": "NON_PARTICIPATING",
|
| 73 |
+
"160": "NON_PARTICIPATING",
|
| 74 |
"161": "NON_PARTICIPATING",
|
| 75 |
"162": "NON_PARTICIPATING",
|
| 76 |
+
"163": "NON_PARTICIPATING",
|
| 77 |
"164": "SUCCESS",
|
| 78 |
"165": "SUCCESS",
|
| 79 |
"166": "NON_PARTICIPATING",
|
| 80 |
+
"167": "NON_PARTICIPATING",
|
| 81 |
+
"168": "SUCCESS",
|
| 82 |
"169": "NON_PARTICIPATING",
|
| 83 |
+
"17": "NOT_ALIVE",
|
| 84 |
"170": "NON_PARTICIPATING",
|
| 85 |
+
"171": "NON_PARTICIPATING",
|
| 86 |
"172": "NON_PARTICIPATING",
|
| 87 |
+
"173": "NOT_ALIVE",
|
| 88 |
"174": "NON_PARTICIPATING",
|
| 89 |
+
"175": "NON_PARTICIPATING",
|
| 90 |
"176": "NOT_ALIVE",
|
| 91 |
"177": "SUCCESS",
|
| 92 |
"178": "NOT_ALIVE",
|
| 93 |
+
"179": "NON_PARTICIPATING",
|
| 94 |
+
"18": "NOT_ALIVE",
|
| 95 |
+
"180": "NON_PARTICIPATING",
|
| 96 |
"181": "SUCCESS",
|
| 97 |
+
"182": "NON_PARTICIPATING",
|
| 98 |
"183": "NON_PARTICIPATING",
|
| 99 |
+
"184": "NOT_ALIVE",
|
| 100 |
"185": "NON_PARTICIPATING",
|
| 101 |
+
"186": "NOT_ALIVE",
|
| 102 |
"187": "NON_PARTICIPATING",
|
| 103 |
+
"188": "NOT_ALIVE",
|
| 104 |
"189": "NOT_ALIVE",
|
| 105 |
"19": "NOT_ALIVE",
|
| 106 |
"190": "NOT_ALIVE",
|
| 107 |
"191": "NOT_ALIVE",
|
| 108 |
"192": "NON_PARTICIPATING",
|
| 109 |
"193": "NON_PARTICIPATING",
|
| 110 |
+
"194": "NOT_ALIVE",
|
| 111 |
"195": "NOT_ALIVE",
|
| 112 |
"196": "NOT_ALIVE",
|
| 113 |
"197": "NON_PARTICIPATING",
|
|
|
|
| 119 |
"201": "NON_PARTICIPATING",
|
| 120 |
"202": "NON_PARTICIPATING",
|
| 121 |
"203": "NON_PARTICIPATING",
|
| 122 |
+
"204": "NOT_ALIVE",
|
| 123 |
"205": "NON_PARTICIPATING",
|
| 124 |
+
"206": "NOT_ALIVE",
|
| 125 |
"207": "NOT_ALIVE",
|
| 126 |
"208": "NOT_ALIVE",
|
| 127 |
"209": "NON_PARTICIPATING",
|
|
|
|
| 129 |
"210": "SUCCESS",
|
| 130 |
"211": "NON_PARTICIPATING",
|
| 131 |
"212": "NOT_ALIVE",
|
| 132 |
+
"213": "SUCCESS",
|
| 133 |
"214": "NON_PARTICIPATING",
|
| 134 |
+
"215": "NOT_ALIVE",
|
| 135 |
+
"216": "NOT_ALIVE",
|
| 136 |
+
"217": "NON_PARTICIPATING",
|
| 137 |
"218": "NON_PARTICIPATING",
|
| 138 |
"219": "NON_PARTICIPATING",
|
| 139 |
"22": "NON_PARTICIPATING",
|
| 140 |
"220": "NON_PARTICIPATING",
|
| 141 |
"221": "SUCCESS",
|
| 142 |
+
"222": "NOT_ALIVE",
|
| 143 |
+
"223": "NON_PARTICIPATING",
|
| 144 |
"224": "NOT_ALIVE",
|
| 145 |
"225": "SUCCESS",
|
| 146 |
+
"226": "NON_PARTICIPATING",
|
| 147 |
"227": "NON_PARTICIPATING",
|
| 148 |
"228": "NON_PARTICIPATING",
|
| 149 |
+
"229": "NON_PARTICIPATING",
|
| 150 |
"23": "NON_PARTICIPATING",
|
| 151 |
+
"230": "NOT_ALIVE",
|
| 152 |
"231": "NON_PARTICIPATING",
|
| 153 |
"232": "NON_PARTICIPATING",
|
| 154 |
+
"233": "NON_PARTICIPATING",
|
| 155 |
"234": "NOT_ALIVE",
|
| 156 |
+
"235": "NON_PARTICIPATING",
|
| 157 |
"236": "NON_PARTICIPATING",
|
| 158 |
+
"237": "NON_PARTICIPATING",
|
| 159 |
"238": "SUCCESS",
|
| 160 |
"239": "NON_PARTICIPATING",
|
| 161 |
"24": "SUCCESS",
|
| 162 |
"240": "NON_PARTICIPATING",
|
| 163 |
+
"241": "NON_PARTICIPATING",
|
| 164 |
"242": "SUCCESS",
|
| 165 |
"243": "NON_PARTICIPATING",
|
| 166 |
"244": "NON_PARTICIPATING",
|
| 167 |
+
"245": "NON_PARTICIPATING",
|
| 168 |
"246": "NON_PARTICIPATING",
|
| 169 |
"247": "NON_PARTICIPATING",
|
| 170 |
"248": "NON_PARTICIPATING",
|
| 171 |
+
"249": "NON_PARTICIPATING",
|
| 172 |
"25": "SUCCESS",
|
| 173 |
"250": "NON_PARTICIPATING",
|
| 174 |
"251": "NOT_ALIVE",
|
|
|
|
| 178 |
"255": "NON_PARTICIPATING",
|
| 179 |
"26": "NON_PARTICIPATING",
|
| 180 |
"27": "SUCCESS",
|
| 181 |
+
"28": "NOT_ALIVE",
|
| 182 |
+
"29": "NON_PARTICIPATING",
|
| 183 |
"3": "NON_PARTICIPATING",
|
| 184 |
"30": "SUCCESS",
|
| 185 |
"31": "NON_PARTICIPATING",
|
| 186 |
"32": "SUCCESS",
|
| 187 |
"33": "NOT_ALIVE",
|
| 188 |
+
"34": "NOT_ALIVE",
|
| 189 |
+
"35": "NOT_ALIVE",
|
| 190 |
"36": "NON_PARTICIPATING",
|
| 191 |
"37": "NON_PARTICIPATING",
|
| 192 |
"38": "SUCCESS",
|
| 193 |
+
"39": "SUCCESS",
|
| 194 |
"4": "NON_PARTICIPATING",
|
| 195 |
"40": "SUCCESS",
|
| 196 |
"41": "SUCCESS",
|
|
|
|
| 209 |
"53": "NOT_ALIVE",
|
| 210 |
"54": "NON_PARTICIPATING",
|
| 211 |
"55": "NON_PARTICIPATING",
|
| 212 |
+
"56": "SUCCESS",
|
| 213 |
+
"57": "NOT_ALIVE",
|
| 214 |
+
"58": "NON_PARTICIPATING",
|
| 215 |
"59": "NON_PARTICIPATING",
|
| 216 |
+
"6": "NON_PARTICIPATING",
|
| 217 |
"60": "NON_PARTICIPATING",
|
| 218 |
"61": "NOT_ALIVE",
|
| 219 |
+
"62": "SUCCESS",
|
| 220 |
"63": "NON_PARTICIPATING",
|
| 221 |
"64": "NON_PARTICIPATING",
|
| 222 |
"65": "NOT_ALIVE",
|
| 223 |
+
"66": "NOT_ALIVE",
|
| 224 |
+
"67": "NOT_ALIVE",
|
| 225 |
"68": "NON_PARTICIPATING",
|
| 226 |
"69": "NON_PARTICIPATING",
|
| 227 |
"7": "NON_PARTICIPATING",
|
| 228 |
+
"70": "NOT_ALIVE",
|
| 229 |
+
"71": "NON_PARTICIPATING",
|
| 230 |
+
"72": "NON_PARTICIPATING",
|
| 231 |
"73": "NON_PARTICIPATING",
|
| 232 |
"74": "NON_PARTICIPATING",
|
| 233 |
"75": "SUCCESS",
|
| 234 |
"76": "NON_PARTICIPATING",
|
| 235 |
"77": "NOT_ALIVE",
|
| 236 |
+
"78": "NOT_ALIVE",
|
| 237 |
"79": "NON_PARTICIPATING",
|
| 238 |
"8": "NON_PARTICIPATING",
|
| 239 |
"80": "NON_PARTICIPATING",
|
| 240 |
"81": "NON_PARTICIPATING",
|
| 241 |
"82": "NOT_ALIVE",
|
| 242 |
"83": "NON_PARTICIPATING",
|
| 243 |
+
"84": "NOT_ALIVE",
|
| 244 |
"85": "NOT_ALIVE",
|
| 245 |
+
"86": "SUCCESS",
|
| 246 |
"87": "NON_PARTICIPATING",
|
| 247 |
+
"88": "NON_PARTICIPATING",
|
| 248 |
"89": "NON_PARTICIPATING",
|
| 249 |
"9": "NON_PARTICIPATING",
|
| 250 |
"90": "SUCCESS",
|
| 251 |
+
"91": "NON_PARTICIPATING",
|
| 252 |
+
"92": "NOT_ALIVE",
|
| 253 |
"93": "NON_PARTICIPATING",
|
| 254 |
"94": "NON_PARTICIPATING",
|
| 255 |
"95": "NON_PARTICIPATING",
|
| 256 |
"96": "NON_PARTICIPATING",
|
| 257 |
"97": "NON_PARTICIPATING",
|
| 258 |
+
"98": "NOT_ALIVE",
|
| 259 |
"99": "NON_PARTICIPATING"
|
| 260 |
},
|
| 261 |
"architectures": [
|
|
|
|
| 264 |
"attention_bias": false,
|
| 265 |
"attention_dropout": 0.0,
|
| 266 |
"block_list": [
|
| 267 |
+
5912018,
|
| 268 |
+
5912028
|
| 269 |
],
|
| 270 |
"bos_token_id": 1,
|
| 271 |
"eos_token_id": 2,
|
| 272 |
"hidden_act": "silu",
|
| 273 |
"hidden_size": 2048,
|
| 274 |
"initializer_range": 0.02,
|
| 275 |
+
"inner_step": 1,
|
| 276 |
"intermediate_size": 5632,
|
| 277 |
+
"last_allreduce_block": 5911865,
|
| 278 |
"max_position_embeddings": 2048,
|
| 279 |
"mlp_bias": false,
|
| 280 |
"model_type": "llama",
|
inner_optimizer.pt
CHANGED
|
@@ -1,3 +1,3 @@
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
-
oid sha256:
|
| 3 |
size 8800564561
|
|
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:fd2cbc1e1f7230e1cca242c3bacbaf5ceba3e487636ac602fb5c1a4ad8c46206
|
| 3 |
size 8800564561
|
model.safetensors
CHANGED
|
@@ -1,3 +1,3 @@
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
-
oid sha256:
|
| 3 |
size 4400216536
|
|
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:d640cdefa2d455f3d473201b23daa225b2cace0c4663cf56f2b3d93c38de869d
|
| 3 |
size 4400216536
|