Run 8. Outer Step 48. Inner Step 6.
Browse files- config.json +102 -103
- gradients.pt +1 -1
- inner_optimizer.pt +1 -1
- model.safetensors +1 -1
config.json
CHANGED
|
@@ -5,141 +5,141 @@
|
|
| 5 |
"1": "NOT_ALIVE",
|
| 6 |
"10": "NOT_ALIVE",
|
| 7 |
"100": "NON_PARTICIPATING",
|
| 8 |
-
"101": "
|
| 9 |
"102": "NON_PARTICIPATING",
|
| 10 |
"103": "NOT_ALIVE",
|
| 11 |
"104": "NON_PARTICIPATING",
|
| 12 |
"105": "NOT_ALIVE",
|
| 13 |
-
"106": "
|
| 14 |
-
"107": "
|
| 15 |
"108": "NOT_ALIVE",
|
| 16 |
"109": "NON_PARTICIPATING",
|
| 17 |
"11": "NOT_ALIVE",
|
| 18 |
-
"110": "
|
| 19 |
"111": "NOT_ALIVE",
|
| 20 |
"112": "SUCCESS",
|
| 21 |
-
"113": "
|
| 22 |
"114": "NOT_ALIVE",
|
| 23 |
-
"115": "
|
| 24 |
-
"116": "
|
| 25 |
-
"117": "
|
| 26 |
-
"118": "
|
| 27 |
"119": "NOT_ALIVE",
|
| 28 |
"12": "NOT_ALIVE",
|
| 29 |
"120": "NOT_ALIVE",
|
| 30 |
-
"121": "
|
| 31 |
-
"122": "
|
| 32 |
-
"123": "
|
| 33 |
"124": "NOT_ALIVE",
|
| 34 |
"125": "NOT_ALIVE",
|
| 35 |
-
"126": "
|
| 36 |
-
"127": "
|
| 37 |
-
"128": "
|
| 38 |
-
"129": "
|
| 39 |
"13": "NOT_ALIVE",
|
| 40 |
"130": "NON_PARTICIPATING",
|
| 41 |
-
"131": "
|
| 42 |
-
"132": "
|
| 43 |
-
"133": "
|
| 44 |
"134": "NON_PARTICIPATING",
|
| 45 |
"135": "SUCCESS",
|
| 46 |
-
"136": "
|
| 47 |
-
"137": "
|
| 48 |
"138": "NON_PARTICIPATING",
|
| 49 |
"139": "NOT_ALIVE",
|
| 50 |
"14": "NOT_ALIVE",
|
| 51 |
-
"140": "
|
| 52 |
"141": "NOT_ALIVE",
|
| 53 |
-
"142": "
|
| 54 |
-
"143": "
|
| 55 |
-
"144": "
|
| 56 |
"145": "SUCCESS",
|
| 57 |
"146": "NON_PARTICIPATING",
|
| 58 |
-
"147": "
|
| 59 |
"148": "NON_PARTICIPATING",
|
| 60 |
-
"149": "
|
| 61 |
"15": "SUCCESS",
|
| 62 |
-
"150": "
|
| 63 |
"151": "NON_PARTICIPATING",
|
| 64 |
-
"152": "
|
| 65 |
-
"153": "
|
| 66 |
"154": "NOT_ALIVE",
|
| 67 |
"155": "FAIL",
|
| 68 |
"156": "NOT_ALIVE",
|
| 69 |
"157": "NON_PARTICIPATING",
|
| 70 |
"158": "NON_PARTICIPATING",
|
| 71 |
-
"159": "
|
| 72 |
"16": "NOT_ALIVE",
|
| 73 |
"160": "NON_PARTICIPATING",
|
| 74 |
-
"161": "
|
| 75 |
"162": "NON_PARTICIPATING",
|
| 76 |
-
"163": "
|
| 77 |
-
"164": "
|
| 78 |
-
"165": "
|
| 79 |
"166": "NOT_ALIVE",
|
| 80 |
-
"167": "
|
| 81 |
"168": "NOT_ALIVE",
|
| 82 |
"169": "SUCCESS",
|
| 83 |
"17": "NOT_ALIVE",
|
| 84 |
-
"170": "
|
| 85 |
"171": "NON_PARTICIPATING",
|
| 86 |
"172": "NON_PARTICIPATING",
|
| 87 |
-
"173": "
|
| 88 |
"174": "SUCCESS",
|
| 89 |
-
"175": "
|
| 90 |
"176": "NOT_ALIVE",
|
| 91 |
"177": "SUCCESS",
|
| 92 |
"178": "NON_PARTICIPATING",
|
| 93 |
-
"179": "
|
| 94 |
"18": "NOT_ALIVE",
|
| 95 |
"180": "NON_PARTICIPATING",
|
| 96 |
-
"181": "
|
| 97 |
"182": "NOT_ALIVE",
|
| 98 |
-
"183": "
|
| 99 |
"184": "NOT_ALIVE",
|
| 100 |
-
"185": "
|
| 101 |
-
"186": "
|
| 102 |
"187": "NOT_ALIVE",
|
| 103 |
"188": "FAIL",
|
| 104 |
"189": "SUCCESS",
|
| 105 |
"19": "NOT_ALIVE",
|
| 106 |
"190": "NON_PARTICIPATING",
|
| 107 |
-
"191": "
|
| 108 |
-
"192": "
|
| 109 |
-
"193": "
|
| 110 |
"194": "NON_PARTICIPATING",
|
| 111 |
"195": "NON_PARTICIPATING",
|
| 112 |
"196": "SUCCESS",
|
| 113 |
"197": "NOT_ALIVE",
|
| 114 |
-
"198": "
|
| 115 |
"199": "SUCCESS",
|
| 116 |
"2": "NOT_ALIVE",
|
| 117 |
"20": "NOT_ALIVE",
|
| 118 |
"200": "NOT_ALIVE",
|
| 119 |
"201": "NOT_ALIVE",
|
| 120 |
"202": "FAIL",
|
| 121 |
-
"203": "
|
| 122 |
-
"204": "
|
| 123 |
-
"205": "
|
| 124 |
"206": "NON_PARTICIPATING",
|
| 125 |
"207": "NOT_ALIVE",
|
| 126 |
"208": "NOT_ALIVE",
|
| 127 |
-
"209": "
|
| 128 |
"21": "NOT_ALIVE",
|
| 129 |
-
"210": "
|
| 130 |
-
"211": "
|
| 131 |
-
"212": "
|
| 132 |
"213": "NON_PARTICIPATING",
|
| 133 |
"214": "NON_PARTICIPATING",
|
| 134 |
"215": "FAIL",
|
| 135 |
"216": "SUCCESS",
|
| 136 |
"217": "SUCCESS",
|
| 137 |
"218": "FAIL",
|
| 138 |
-
"219": "
|
| 139 |
"22": "NOT_ALIVE",
|
| 140 |
-
"220": "
|
| 141 |
"221": "SUCCESS",
|
| 142 |
-
"222": "
|
| 143 |
"223": "NOT_ALIVE",
|
| 144 |
"224": "NON_PARTICIPATING",
|
| 145 |
"225": "SUCCESS",
|
|
@@ -148,7 +148,7 @@
|
|
| 148 |
"228": "SUCCESS",
|
| 149 |
"229": "NON_PARTICIPATING",
|
| 150 |
"23": "NOT_ALIVE",
|
| 151 |
-
"230": "
|
| 152 |
"231": "NOT_ALIVE",
|
| 153 |
"232": "NOT_ALIVE",
|
| 154 |
"233": "NON_PARTICIPATING",
|
|
@@ -157,105 +157,105 @@
|
|
| 157 |
"236": "SUCCESS",
|
| 158 |
"237": "NON_PARTICIPATING",
|
| 159 |
"238": "SUCCESS",
|
| 160 |
-
"239": "
|
| 161 |
-
"24": "
|
| 162 |
"240": "NON_PARTICIPATING",
|
| 163 |
-
"241": "
|
| 164 |
"242": "NON_PARTICIPATING",
|
| 165 |
"243": "NON_PARTICIPATING",
|
| 166 |
"244": "NON_PARTICIPATING",
|
| 167 |
"245": "NON_PARTICIPATING",
|
| 168 |
"246": "NON_PARTICIPATING",
|
| 169 |
-
"247": "
|
| 170 |
-
"248": "
|
| 171 |
-
"249": "
|
| 172 |
"25": "SUCCESS",
|
| 173 |
-
"250": "
|
| 174 |
-
"251": "
|
| 175 |
-
"252": "
|
| 176 |
"253": "SUCCESS",
|
| 177 |
-
"254": "
|
| 178 |
"255": "NON_PARTICIPATING",
|
| 179 |
-
"26": "
|
| 180 |
"27": "SUCCESS",
|
| 181 |
"28": "NON_PARTICIPATING",
|
| 182 |
-
"29": "
|
| 183 |
"3": "FAIL",
|
| 184 |
"30": "SUCCESS",
|
| 185 |
-
"31": "
|
| 186 |
-
"32": "
|
| 187 |
"33": "NOT_ALIVE",
|
| 188 |
"34": "NOT_ALIVE",
|
| 189 |
-
"35": "
|
| 190 |
-
"36": "
|
| 191 |
"37": "NOT_ALIVE",
|
| 192 |
"38": "SUCCESS",
|
| 193 |
-
"39": "
|
| 194 |
"4": "NOT_ALIVE",
|
| 195 |
"40": "NON_PARTICIPATING",
|
| 196 |
"41": "FAIL",
|
| 197 |
"42": "NOT_ALIVE",
|
| 198 |
"43": "NON_PARTICIPATING",
|
| 199 |
-
"44": "
|
| 200 |
"45": "NOT_ALIVE",
|
| 201 |
"46": "NON_PARTICIPATING",
|
| 202 |
"47": "NOT_ALIVE",
|
| 203 |
-
"48": "
|
| 204 |
"49": "NON_PARTICIPATING",
|
| 205 |
-
"5": "
|
| 206 |
"50": "NOT_ALIVE",
|
| 207 |
-
"51": "
|
| 208 |
"52": "SUCCESS",
|
| 209 |
"53": "SUCCESS",
|
| 210 |
-
"54": "
|
| 211 |
"55": "NON_PARTICIPATING",
|
| 212 |
"56": "NOT_ALIVE",
|
| 213 |
-
"57": "
|
| 214 |
"58": "NOT_ALIVE",
|
| 215 |
"59": "NOT_ALIVE",
|
| 216 |
"6": "NOT_ALIVE",
|
| 217 |
-
"60": "
|
| 218 |
"61": "NON_PARTICIPATING",
|
| 219 |
"62": "NOT_ALIVE",
|
| 220 |
-
"63": "
|
| 221 |
"64": "NOT_ALIVE",
|
| 222 |
"65": "NOT_ALIVE",
|
| 223 |
"66": "NON_PARTICIPATING",
|
| 224 |
"67": "NON_PARTICIPATING",
|
| 225 |
-
"68": "
|
| 226 |
"69": "NON_PARTICIPATING",
|
| 227 |
"7": "NOT_ALIVE",
|
| 228 |
"70": "NOT_ALIVE",
|
| 229 |
-
"71": "
|
| 230 |
-
"72": "
|
| 231 |
"73": "NOT_ALIVE",
|
| 232 |
-
"74": "
|
| 233 |
"75": "NOT_ALIVE",
|
| 234 |
-
"76": "
|
| 235 |
"77": "NON_PARTICIPATING",
|
| 236 |
"78": "NON_PARTICIPATING",
|
| 237 |
-
"79": "
|
| 238 |
"8": "NOT_ALIVE",
|
| 239 |
-
"80": "
|
| 240 |
"81": "NOT_ALIVE",
|
| 241 |
"82": "NON_PARTICIPATING",
|
| 242 |
-
"83": "
|
| 243 |
-
"84": "
|
| 244 |
"85": "NOT_ALIVE",
|
| 245 |
"86": "NOT_ALIVE",
|
| 246 |
-
"87": "
|
| 247 |
"88": "NON_PARTICIPATING",
|
| 248 |
"89": "NON_PARTICIPATING",
|
| 249 |
"9": "NOT_ALIVE",
|
| 250 |
-
"90": "
|
| 251 |
"91": "NON_PARTICIPATING",
|
| 252 |
-
"92": "
|
| 253 |
"93": "NOT_ALIVE",
|
| 254 |
"94": "NOT_ALIVE",
|
| 255 |
"95": "NOT_ALIVE",
|
| 256 |
"96": "NOT_ALIVE",
|
| 257 |
-
"97": "
|
| 258 |
-
"98": "
|
| 259 |
"99": "NON_PARTICIPATING"
|
| 260 |
},
|
| 261 |
"architectures": [
|
|
@@ -264,18 +264,17 @@
|
|
| 264 |
"attention_bias": false,
|
| 265 |
"attention_dropout": 0.0,
|
| 266 |
"block_list": [
|
| 267 |
-
|
| 268 |
-
|
| 269 |
-
6368388
|
| 270 |
],
|
| 271 |
"bos_token_id": 1,
|
| 272 |
"eos_token_id": 2,
|
| 273 |
"hidden_act": "silu",
|
| 274 |
"hidden_size": 2048,
|
| 275 |
"initializer_range": 0.02,
|
| 276 |
-
"inner_step":
|
| 277 |
"intermediate_size": 5632,
|
| 278 |
-
"last_allreduce_block":
|
| 279 |
"max_position_embeddings": 2048,
|
| 280 |
"mlp_bias": false,
|
| 281 |
"model_type": "llama",
|
|
|
|
| 5 |
"1": "NOT_ALIVE",
|
| 6 |
"10": "NOT_ALIVE",
|
| 7 |
"100": "NON_PARTICIPATING",
|
| 8 |
+
"101": "NOT_ALIVE",
|
| 9 |
"102": "NON_PARTICIPATING",
|
| 10 |
"103": "NOT_ALIVE",
|
| 11 |
"104": "NON_PARTICIPATING",
|
| 12 |
"105": "NOT_ALIVE",
|
| 13 |
+
"106": "NOT_ALIVE",
|
| 14 |
+
"107": "NON_PARTICIPATING",
|
| 15 |
"108": "NOT_ALIVE",
|
| 16 |
"109": "NON_PARTICIPATING",
|
| 17 |
"11": "NOT_ALIVE",
|
| 18 |
+
"110": "NON_PARTICIPATING",
|
| 19 |
"111": "NOT_ALIVE",
|
| 20 |
"112": "SUCCESS",
|
| 21 |
+
"113": "NOT_ALIVE",
|
| 22 |
"114": "NOT_ALIVE",
|
| 23 |
+
"115": "NOT_ALIVE",
|
| 24 |
+
"116": "NOT_ALIVE",
|
| 25 |
+
"117": "NOT_ALIVE",
|
| 26 |
+
"118": "NOT_ALIVE",
|
| 27 |
"119": "NOT_ALIVE",
|
| 28 |
"12": "NOT_ALIVE",
|
| 29 |
"120": "NOT_ALIVE",
|
| 30 |
+
"121": "SUCCESS",
|
| 31 |
+
"122": "NON_PARTICIPATING",
|
| 32 |
+
"123": "SUCCESS",
|
| 33 |
"124": "NOT_ALIVE",
|
| 34 |
"125": "NOT_ALIVE",
|
| 35 |
+
"126": "FAIL",
|
| 36 |
+
"127": "NON_PARTICIPATING",
|
| 37 |
+
"128": "SUCCESS",
|
| 38 |
+
"129": "NON_PARTICIPATING",
|
| 39 |
"13": "NOT_ALIVE",
|
| 40 |
"130": "NON_PARTICIPATING",
|
| 41 |
+
"131": "NOT_ALIVE",
|
| 42 |
+
"132": "SUCCESS",
|
| 43 |
+
"133": "SUCCESS",
|
| 44 |
"134": "NON_PARTICIPATING",
|
| 45 |
"135": "SUCCESS",
|
| 46 |
+
"136": "SUCCESS",
|
| 47 |
+
"137": "FAIL",
|
| 48 |
"138": "NON_PARTICIPATING",
|
| 49 |
"139": "NOT_ALIVE",
|
| 50 |
"14": "NOT_ALIVE",
|
| 51 |
+
"140": "NON_PARTICIPATING",
|
| 52 |
"141": "NOT_ALIVE",
|
| 53 |
+
"142": "NON_PARTICIPATING",
|
| 54 |
+
"143": "NOT_ALIVE",
|
| 55 |
+
"144": "NON_PARTICIPATING",
|
| 56 |
"145": "SUCCESS",
|
| 57 |
"146": "NON_PARTICIPATING",
|
| 58 |
+
"147": "NOT_ALIVE",
|
| 59 |
"148": "NON_PARTICIPATING",
|
| 60 |
+
"149": "SUCCESS",
|
| 61 |
"15": "SUCCESS",
|
| 62 |
+
"150": "SUCCESS",
|
| 63 |
"151": "NON_PARTICIPATING",
|
| 64 |
+
"152": "FAIL",
|
| 65 |
+
"153": "SUCCESS",
|
| 66 |
"154": "NOT_ALIVE",
|
| 67 |
"155": "FAIL",
|
| 68 |
"156": "NOT_ALIVE",
|
| 69 |
"157": "NON_PARTICIPATING",
|
| 70 |
"158": "NON_PARTICIPATING",
|
| 71 |
+
"159": "NOT_ALIVE",
|
| 72 |
"16": "NOT_ALIVE",
|
| 73 |
"160": "NON_PARTICIPATING",
|
| 74 |
+
"161": "NON_PARTICIPATING",
|
| 75 |
"162": "NON_PARTICIPATING",
|
| 76 |
+
"163": "NOT_ALIVE",
|
| 77 |
+
"164": "FAIL",
|
| 78 |
+
"165": "SUCCESS",
|
| 79 |
"166": "NOT_ALIVE",
|
| 80 |
+
"167": "NOT_ALIVE",
|
| 81 |
"168": "NOT_ALIVE",
|
| 82 |
"169": "SUCCESS",
|
| 83 |
"17": "NOT_ALIVE",
|
| 84 |
+
"170": "SUCCESS",
|
| 85 |
"171": "NON_PARTICIPATING",
|
| 86 |
"172": "NON_PARTICIPATING",
|
| 87 |
+
"173": "NOT_ALIVE",
|
| 88 |
"174": "SUCCESS",
|
| 89 |
+
"175": "NON_PARTICIPATING",
|
| 90 |
"176": "NOT_ALIVE",
|
| 91 |
"177": "SUCCESS",
|
| 92 |
"178": "NON_PARTICIPATING",
|
| 93 |
+
"179": "SUCCESS",
|
| 94 |
"18": "NOT_ALIVE",
|
| 95 |
"180": "NON_PARTICIPATING",
|
| 96 |
+
"181": "FAIL",
|
| 97 |
"182": "NOT_ALIVE",
|
| 98 |
+
"183": "NOT_ALIVE",
|
| 99 |
"184": "NOT_ALIVE",
|
| 100 |
+
"185": "NON_PARTICIPATING",
|
| 101 |
+
"186": "FAIL",
|
| 102 |
"187": "NOT_ALIVE",
|
| 103 |
"188": "FAIL",
|
| 104 |
"189": "SUCCESS",
|
| 105 |
"19": "NOT_ALIVE",
|
| 106 |
"190": "NON_PARTICIPATING",
|
| 107 |
+
"191": "SUCCESS",
|
| 108 |
+
"192": "NON_PARTICIPATING",
|
| 109 |
+
"193": "NOT_ALIVE",
|
| 110 |
"194": "NON_PARTICIPATING",
|
| 111 |
"195": "NON_PARTICIPATING",
|
| 112 |
"196": "SUCCESS",
|
| 113 |
"197": "NOT_ALIVE",
|
| 114 |
+
"198": "NOT_ALIVE",
|
| 115 |
"199": "SUCCESS",
|
| 116 |
"2": "NOT_ALIVE",
|
| 117 |
"20": "NOT_ALIVE",
|
| 118 |
"200": "NOT_ALIVE",
|
| 119 |
"201": "NOT_ALIVE",
|
| 120 |
"202": "FAIL",
|
| 121 |
+
"203": "NOT_ALIVE",
|
| 122 |
+
"204": "NON_PARTICIPATING",
|
| 123 |
+
"205": "NOT_ALIVE",
|
| 124 |
"206": "NON_PARTICIPATING",
|
| 125 |
"207": "NOT_ALIVE",
|
| 126 |
"208": "NOT_ALIVE",
|
| 127 |
+
"209": "NOT_ALIVE",
|
| 128 |
"21": "NOT_ALIVE",
|
| 129 |
+
"210": "NOT_ALIVE",
|
| 130 |
+
"211": "SUCCESS",
|
| 131 |
+
"212": "NOT_ALIVE",
|
| 132 |
"213": "NON_PARTICIPATING",
|
| 133 |
"214": "NON_PARTICIPATING",
|
| 134 |
"215": "FAIL",
|
| 135 |
"216": "SUCCESS",
|
| 136 |
"217": "SUCCESS",
|
| 137 |
"218": "FAIL",
|
| 138 |
+
"219": "FAIL",
|
| 139 |
"22": "NOT_ALIVE",
|
| 140 |
+
"220": "FAIL",
|
| 141 |
"221": "SUCCESS",
|
| 142 |
+
"222": "NOT_ALIVE",
|
| 143 |
"223": "NOT_ALIVE",
|
| 144 |
"224": "NON_PARTICIPATING",
|
| 145 |
"225": "SUCCESS",
|
|
|
|
| 148 |
"228": "SUCCESS",
|
| 149 |
"229": "NON_PARTICIPATING",
|
| 150 |
"23": "NOT_ALIVE",
|
| 151 |
+
"230": "NOT_ALIVE",
|
| 152 |
"231": "NOT_ALIVE",
|
| 153 |
"232": "NOT_ALIVE",
|
| 154 |
"233": "NON_PARTICIPATING",
|
|
|
|
| 157 |
"236": "SUCCESS",
|
| 158 |
"237": "NON_PARTICIPATING",
|
| 159 |
"238": "SUCCESS",
|
| 160 |
+
"239": "NON_PARTICIPATING",
|
| 161 |
+
"24": "SUCCESS",
|
| 162 |
"240": "NON_PARTICIPATING",
|
| 163 |
+
"241": "NON_PARTICIPATING",
|
| 164 |
"242": "NON_PARTICIPATING",
|
| 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": "NOT_ALIVE",
|
| 172 |
"25": "SUCCESS",
|
| 173 |
+
"250": "NON_PARTICIPATING",
|
| 174 |
+
"251": "NON_PARTICIPATING",
|
| 175 |
+
"252": "SUCCESS",
|
| 176 |
"253": "SUCCESS",
|
| 177 |
+
"254": "NOT_ALIVE",
|
| 178 |
"255": "NON_PARTICIPATING",
|
| 179 |
+
"26": "FAIL",
|
| 180 |
"27": "SUCCESS",
|
| 181 |
"28": "NON_PARTICIPATING",
|
| 182 |
+
"29": "SUCCESS",
|
| 183 |
"3": "FAIL",
|
| 184 |
"30": "SUCCESS",
|
| 185 |
+
"31": "NON_PARTICIPATING",
|
| 186 |
+
"32": "NON_PARTICIPATING",
|
| 187 |
"33": "NOT_ALIVE",
|
| 188 |
"34": "NOT_ALIVE",
|
| 189 |
+
"35": "NON_PARTICIPATING",
|
| 190 |
+
"36": "FAIL",
|
| 191 |
"37": "NOT_ALIVE",
|
| 192 |
"38": "SUCCESS",
|
| 193 |
+
"39": "NOT_ALIVE",
|
| 194 |
"4": "NOT_ALIVE",
|
| 195 |
"40": "NON_PARTICIPATING",
|
| 196 |
"41": "FAIL",
|
| 197 |
"42": "NOT_ALIVE",
|
| 198 |
"43": "NON_PARTICIPATING",
|
| 199 |
+
"44": "NON_PARTICIPATING",
|
| 200 |
"45": "NOT_ALIVE",
|
| 201 |
"46": "NON_PARTICIPATING",
|
| 202 |
"47": "NOT_ALIVE",
|
| 203 |
+
"48": "NON_PARTICIPATING",
|
| 204 |
"49": "NON_PARTICIPATING",
|
| 205 |
+
"5": "NOT_ALIVE",
|
| 206 |
"50": "NOT_ALIVE",
|
| 207 |
+
"51": "NON_PARTICIPATING",
|
| 208 |
"52": "SUCCESS",
|
| 209 |
"53": "SUCCESS",
|
| 210 |
+
"54": "NON_PARTICIPATING",
|
| 211 |
"55": "NON_PARTICIPATING",
|
| 212 |
"56": "NOT_ALIVE",
|
| 213 |
+
"57": "NOT_ALIVE",
|
| 214 |
"58": "NOT_ALIVE",
|
| 215 |
"59": "NOT_ALIVE",
|
| 216 |
"6": "NOT_ALIVE",
|
| 217 |
+
"60": "NOT_ALIVE",
|
| 218 |
"61": "NON_PARTICIPATING",
|
| 219 |
"62": "NOT_ALIVE",
|
| 220 |
+
"63": "FAIL",
|
| 221 |
"64": "NOT_ALIVE",
|
| 222 |
"65": "NOT_ALIVE",
|
| 223 |
"66": "NON_PARTICIPATING",
|
| 224 |
"67": "NON_PARTICIPATING",
|
| 225 |
+
"68": "NOT_ALIVE",
|
| 226 |
"69": "NON_PARTICIPATING",
|
| 227 |
"7": "NOT_ALIVE",
|
| 228 |
"70": "NOT_ALIVE",
|
| 229 |
+
"71": "NOT_ALIVE",
|
| 230 |
+
"72": "NOT_ALIVE",
|
| 231 |
"73": "NOT_ALIVE",
|
| 232 |
+
"74": "SUCCESS",
|
| 233 |
"75": "NOT_ALIVE",
|
| 234 |
+
"76": "NOT_ALIVE",
|
| 235 |
"77": "NON_PARTICIPATING",
|
| 236 |
"78": "NON_PARTICIPATING",
|
| 237 |
+
"79": "SUCCESS",
|
| 238 |
"8": "NOT_ALIVE",
|
| 239 |
+
"80": "NON_PARTICIPATING",
|
| 240 |
"81": "NOT_ALIVE",
|
| 241 |
"82": "NON_PARTICIPATING",
|
| 242 |
+
"83": "NOT_ALIVE",
|
| 243 |
+
"84": "NON_PARTICIPATING",
|
| 244 |
"85": "NOT_ALIVE",
|
| 245 |
"86": "NOT_ALIVE",
|
| 246 |
+
"87": "FAIL",
|
| 247 |
"88": "NON_PARTICIPATING",
|
| 248 |
"89": "NON_PARTICIPATING",
|
| 249 |
"9": "NOT_ALIVE",
|
| 250 |
+
"90": "NON_PARTICIPATING",
|
| 251 |
"91": "NON_PARTICIPATING",
|
| 252 |
+
"92": "NOT_ALIVE",
|
| 253 |
"93": "NOT_ALIVE",
|
| 254 |
"94": "NOT_ALIVE",
|
| 255 |
"95": "NOT_ALIVE",
|
| 256 |
"96": "NOT_ALIVE",
|
| 257 |
+
"97": "FAIL",
|
| 258 |
+
"98": "NON_PARTICIPATING",
|
| 259 |
"99": "NON_PARTICIPATING"
|
| 260 |
},
|
| 261 |
"architectures": [
|
|
|
|
| 264 |
"attention_bias": false,
|
| 265 |
"attention_dropout": 0.0,
|
| 266 |
"block_list": [
|
| 267 |
+
6373306,
|
| 268 |
+
6373309
|
|
|
|
| 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": 5,
|
| 276 |
"intermediate_size": 5632,
|
| 277 |
+
"last_allreduce_block": 6372758,
|
| 278 |
"max_position_embeddings": 2048,
|
| 279 |
"mlp_bias": false,
|
| 280 |
"model_type": "llama",
|
gradients.pt
CHANGED
|
@@ -1,3 +1,3 @@
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
-
oid sha256:
|
| 3 |
size 88389626
|
|
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:69c714f8b67d397fabc46588855da1999a7e99c1833766fa0fe6eefca4907168
|
| 3 |
size 88389626
|
inner_optimizer.pt
CHANGED
|
@@ -1,3 +1,3 @@
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
-
oid sha256:
|
| 3 |
size 4924926906
|
|
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:d1d60dda281d63cdb77cb64908ad541484b90208455167b93382e3cc598473cc
|
| 3 |
size 4924926906
|
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:cde46d137120d6bb8b5be88f09116fd540a71259b44cd506358a0ab5e9525bde
|
| 3 |
size 4400216536
|