Run 8. Outer Step 84. Inner Step 5.
Browse files- config.json +102 -102
- gradients.pt +1 -1
- inner_optimizer.pt +1 -1
- model.safetensors +1 -1
config.json
CHANGED
|
@@ -6,140 +6,140 @@
|
|
| 6 |
"10": "NOT_ALIVE",
|
| 7 |
"100": "NON_PARTICIPATING",
|
| 8 |
"101": "NON_PARTICIPATING",
|
| 9 |
-
"102": "
|
| 10 |
-
"103": "
|
| 11 |
"104": "NON_PARTICIPATING",
|
| 12 |
-
"105": "
|
| 13 |
-
"106": "
|
| 14 |
"107": "NON_PARTICIPATING",
|
| 15 |
"108": "NON_PARTICIPATING",
|
| 16 |
"109": "SUCCESS",
|
| 17 |
"11": "NOT_ALIVE",
|
| 18 |
-
"110": "
|
| 19 |
-
"111": "
|
| 20 |
"112": "NON_PARTICIPATING",
|
| 21 |
"113": "NOT_ALIVE",
|
| 22 |
"114": "NON_PARTICIPATING",
|
| 23 |
-
"115": "
|
| 24 |
"116": "NOT_ALIVE",
|
| 25 |
"117": "SUCCESS",
|
| 26 |
-
"118": "
|
| 27 |
-
"119": "
|
| 28 |
"12": "NOT_ALIVE",
|
| 29 |
-
"120": "
|
| 30 |
"121": "NON_PARTICIPATING",
|
| 31 |
"122": "SUCCESS",
|
| 32 |
-
"123": "
|
| 33 |
-
"124": "
|
| 34 |
"125": "NOT_ALIVE",
|
| 35 |
"126": "FAIL",
|
| 36 |
"127": "SUCCESS",
|
| 37 |
"128": "NON_PARTICIPATING",
|
| 38 |
-
"129": "
|
| 39 |
"13": "SUCCESS",
|
| 40 |
"130": "SUCCESS",
|
| 41 |
-
"131": "
|
| 42 |
"132": "SUCCESS",
|
| 43 |
-
"133": "
|
| 44 |
"134": "NON_PARTICIPATING",
|
| 45 |
"135": "SUCCESS",
|
| 46 |
"136": "NON_PARTICIPATING",
|
| 47 |
-
"137": "
|
| 48 |
"138": "NOT_ALIVE",
|
| 49 |
"139": "SUCCESS",
|
| 50 |
"14": "NOT_ALIVE",
|
| 51 |
"140": "SUCCESS",
|
| 52 |
-
"141": "
|
| 53 |
"142": "SUCCESS",
|
| 54 |
"143": "SUCCESS",
|
| 55 |
-
"144": "
|
| 56 |
-
"145": "
|
| 57 |
"146": "NOT_ALIVE",
|
| 58 |
"147": "NON_PARTICIPATING",
|
| 59 |
"148": "NON_PARTICIPATING",
|
| 60 |
-
"149": "
|
| 61 |
"15": "NOT_ALIVE",
|
| 62 |
"150": "SUCCESS",
|
| 63 |
"151": "NON_PARTICIPATING",
|
| 64 |
"152": "NON_PARTICIPATING",
|
| 65 |
"153": "NOT_ALIVE",
|
| 66 |
"154": "SUCCESS",
|
| 67 |
-
"155": "
|
| 68 |
"156": "NOT_ALIVE",
|
| 69 |
-
"157": "
|
| 70 |
"158": "NON_PARTICIPATING",
|
| 71 |
"159": "SUCCESS",
|
| 72 |
-
"16": "
|
| 73 |
"160": "NON_PARTICIPATING",
|
| 74 |
-
"161": "
|
| 75 |
-
"162": "
|
| 76 |
-
"163": "
|
| 77 |
"164": "NON_PARTICIPATING",
|
| 78 |
-
"165": "
|
| 79 |
-
"166": "
|
| 80 |
-
"167": "
|
| 81 |
"168": "NON_PARTICIPATING",
|
| 82 |
"169": "NON_PARTICIPATING",
|
| 83 |
-
"17": "
|
| 84 |
-
"170": "
|
| 85 |
-
"171": "
|
| 86 |
"172": "NOT_ALIVE",
|
| 87 |
"173": "NON_PARTICIPATING",
|
| 88 |
"174": "SUCCESS",
|
| 89 |
"175": "NON_PARTICIPATING",
|
| 90 |
"176": "NOT_ALIVE",
|
| 91 |
-
"177": "
|
| 92 |
-
"178": "
|
| 93 |
-
"179": "
|
| 94 |
-
"18": "
|
| 95 |
-
"180": "
|
| 96 |
"181": "NON_PARTICIPATING",
|
| 97 |
-
"182": "
|
| 98 |
-
"183": "
|
| 99 |
"184": "SUCCESS",
|
| 100 |
-
"185": "
|
| 101 |
-
"186": "
|
| 102 |
"187": "NON_PARTICIPATING",
|
| 103 |
"188": "SUCCESS",
|
| 104 |
-
"189": "
|
| 105 |
"19": "SUCCESS",
|
| 106 |
-
"190": "
|
| 107 |
-
"191": "
|
| 108 |
"192": "SUCCESS",
|
| 109 |
"193": "SUCCESS",
|
| 110 |
"194": "NOT_ALIVE",
|
| 111 |
-
"195": "
|
| 112 |
-
"196": "
|
| 113 |
"197": "NOT_ALIVE",
|
| 114 |
-
"198": "
|
| 115 |
"199": "FAIL",
|
| 116 |
"2": "SUCCESS",
|
| 117 |
"20": "NOT_ALIVE",
|
| 118 |
"200": "SUCCESS",
|
| 119 |
"201": "NOT_ALIVE",
|
| 120 |
-
"202": "
|
| 121 |
-
"203": "
|
| 122 |
"204": "NON_PARTICIPATING",
|
| 123 |
"205": "NON_PARTICIPATING",
|
| 124 |
"206": "SUCCESS",
|
| 125 |
"207": "NOT_ALIVE",
|
| 126 |
-
"208": "
|
| 127 |
"209": "NON_PARTICIPATING",
|
| 128 |
"21": "NOT_ALIVE",
|
| 129 |
-
"210": "
|
| 130 |
"211": "SUCCESS",
|
| 131 |
-
"212": "
|
| 132 |
-
"213": "
|
| 133 |
-
"214": "
|
| 134 |
"215": "SUCCESS",
|
| 135 |
-
"216": "
|
| 136 |
"217": "SUCCESS",
|
| 137 |
"218": "SUCCESS",
|
| 138 |
"219": "SUCCESS",
|
| 139 |
-
"22": "
|
| 140 |
-
"220": "
|
| 141 |
-
"221": "
|
| 142 |
-
"222": "
|
| 143 |
"223": "NON_PARTICIPATING",
|
| 144 |
"224": "NON_PARTICIPATING",
|
| 145 |
"225": "NON_PARTICIPATING",
|
|
@@ -151,23 +151,23 @@
|
|
| 151 |
"230": "NOT_ALIVE",
|
| 152 |
"231": "NON_PARTICIPATING",
|
| 153 |
"232": "NOT_ALIVE",
|
| 154 |
-
"233": "
|
| 155 |
-
"234": "
|
| 156 |
"235": "NON_PARTICIPATING",
|
| 157 |
"236": "SUCCESS",
|
| 158 |
-
"237": "
|
| 159 |
"238": "NON_PARTICIPATING",
|
| 160 |
"239": "NOT_ALIVE",
|
| 161 |
-
"24": "
|
| 162 |
-
"240": "
|
| 163 |
"241": "NON_PARTICIPATING",
|
| 164 |
-
"242": "
|
| 165 |
-
"243": "
|
| 166 |
"244": "SUCCESS",
|
| 167 |
"245": "SUCCESS",
|
| 168 |
"246": "SUCCESS",
|
| 169 |
"247": "FAIL",
|
| 170 |
-
"248": "
|
| 171 |
"249": "SUCCESS",
|
| 172 |
"25": "SUCCESS",
|
| 173 |
"250": "FAIL",
|
|
@@ -175,7 +175,7 @@
|
|
| 175 |
"252": "SUCCESS",
|
| 176 |
"253": "SUCCESS",
|
| 177 |
"254": "NON_PARTICIPATING",
|
| 178 |
-
"255": "
|
| 179 |
"26": "FAIL",
|
| 180 |
"27": "SUCCESS",
|
| 181 |
"28": "NON_PARTICIPATING",
|
|
@@ -185,46 +185,46 @@
|
|
| 185 |
"31": "SUCCESS",
|
| 186 |
"32": "SUCCESS",
|
| 187 |
"33": "NON_PARTICIPATING",
|
| 188 |
-
"34": "
|
| 189 |
"35": "FAIL",
|
| 190 |
-
"36": "
|
| 191 |
-
"37": "
|
| 192 |
-
"38": "
|
| 193 |
"39": "NOT_ALIVE",
|
| 194 |
"4": "NOT_ALIVE",
|
| 195 |
-
"40": "
|
| 196 |
"41": "FAIL",
|
| 197 |
-
"42": "
|
| 198 |
-
"43": "
|
| 199 |
-
"44": "
|
| 200 |
"45": "NON_PARTICIPATING",
|
| 201 |
"46": "NOT_ALIVE",
|
| 202 |
-
"47": "
|
| 203 |
-
"48": "
|
| 204 |
"49": "NON_PARTICIPATING",
|
| 205 |
"5": "NOT_ALIVE",
|
| 206 |
"50": "NOT_ALIVE",
|
| 207 |
-
"51": "
|
| 208 |
"52": "SUCCESS",
|
| 209 |
"53": "NOT_ALIVE",
|
| 210 |
-
"54": "
|
| 211 |
"55": "SUCCESS",
|
| 212 |
"56": "FAIL",
|
| 213 |
"57": "SUCCESS",
|
| 214 |
"58": "SUCCESS",
|
| 215 |
-
"59": "
|
| 216 |
"6": "NOT_ALIVE",
|
| 217 |
-
"60": "
|
| 218 |
-
"61": "
|
| 219 |
-
"62": "
|
| 220 |
-
"63": "
|
| 221 |
"64": "NON_PARTICIPATING",
|
| 222 |
"65": "SUCCESS",
|
| 223 |
"66": "NON_PARTICIPATING",
|
| 224 |
-
"67": "
|
| 225 |
-
"68": "
|
| 226 |
"69": "SUCCESS",
|
| 227 |
-
"7": "
|
| 228 |
"70": "NOT_ALIVE",
|
| 229 |
"71": "SUCCESS",
|
| 230 |
"72": "NON_PARTICIPATING",
|
|
@@ -232,27 +232,27 @@
|
|
| 232 |
"74": "SUCCESS",
|
| 233 |
"75": "NOT_ALIVE",
|
| 234 |
"76": "SUCCESS",
|
| 235 |
-
"77": "
|
| 236 |
"78": "NON_PARTICIPATING",
|
| 237 |
"79": "SUCCESS",
|
| 238 |
"8": "SUCCESS",
|
| 239 |
"80": "NON_PARTICIPATING",
|
| 240 |
"81": "NON_PARTICIPATING",
|
| 241 |
-
"82": "
|
| 242 |
"83": "SUCCESS",
|
| 243 |
"84": "SUCCESS",
|
| 244 |
-
"85": "
|
| 245 |
"86": "NON_PARTICIPATING",
|
| 246 |
"87": "SUCCESS",
|
| 247 |
-
"88": "
|
| 248 |
-
"89": "
|
| 249 |
"9": "NOT_ALIVE",
|
| 250 |
-
"90": "
|
| 251 |
-
"91": "
|
| 252 |
"92": "SUCCESS",
|
| 253 |
-
"93": "
|
| 254 |
-
"94": "
|
| 255 |
-
"95": "
|
| 256 |
"96": "SUCCESS",
|
| 257 |
"97": "SUCCESS",
|
| 258 |
"98": "NOT_ALIVE",
|
|
@@ -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",
|
|
|
|
| 6 |
"10": "NOT_ALIVE",
|
| 7 |
"100": "NON_PARTICIPATING",
|
| 8 |
"101": "NON_PARTICIPATING",
|
| 9 |
+
"102": "FAIL",
|
| 10 |
+
"103": "NOT_ALIVE",
|
| 11 |
"104": "NON_PARTICIPATING",
|
| 12 |
+
"105": "SUCCESS",
|
| 13 |
+
"106": "FAIL",
|
| 14 |
"107": "NON_PARTICIPATING",
|
| 15 |
"108": "NON_PARTICIPATING",
|
| 16 |
"109": "SUCCESS",
|
| 17 |
"11": "NOT_ALIVE",
|
| 18 |
+
"110": "NON_PARTICIPATING",
|
| 19 |
+
"111": "NOT_ALIVE",
|
| 20 |
"112": "NON_PARTICIPATING",
|
| 21 |
"113": "NOT_ALIVE",
|
| 22 |
"114": "NON_PARTICIPATING",
|
| 23 |
+
"115": "SUCCESS",
|
| 24 |
"116": "NOT_ALIVE",
|
| 25 |
"117": "SUCCESS",
|
| 26 |
+
"118": "NOT_ALIVE",
|
| 27 |
+
"119": "NON_PARTICIPATING",
|
| 28 |
"12": "NOT_ALIVE",
|
| 29 |
+
"120": "SUCCESS",
|
| 30 |
"121": "NON_PARTICIPATING",
|
| 31 |
"122": "SUCCESS",
|
| 32 |
+
"123": "FAIL",
|
| 33 |
+
"124": "NOT_ALIVE",
|
| 34 |
"125": "NOT_ALIVE",
|
| 35 |
"126": "FAIL",
|
| 36 |
"127": "SUCCESS",
|
| 37 |
"128": "NON_PARTICIPATING",
|
| 38 |
+
"129": "NON_PARTICIPATING",
|
| 39 |
"13": "SUCCESS",
|
| 40 |
"130": "SUCCESS",
|
| 41 |
+
"131": "NON_PARTICIPATING",
|
| 42 |
"132": "SUCCESS",
|
| 43 |
+
"133": "NOT_ALIVE",
|
| 44 |
"134": "NON_PARTICIPATING",
|
| 45 |
"135": "SUCCESS",
|
| 46 |
"136": "NON_PARTICIPATING",
|
| 47 |
+
"137": "FAIL",
|
| 48 |
"138": "NOT_ALIVE",
|
| 49 |
"139": "SUCCESS",
|
| 50 |
"14": "NOT_ALIVE",
|
| 51 |
"140": "SUCCESS",
|
| 52 |
+
"141": "NOT_ALIVE",
|
| 53 |
"142": "SUCCESS",
|
| 54 |
"143": "SUCCESS",
|
| 55 |
+
"144": "NOT_ALIVE",
|
| 56 |
+
"145": "SUCCESS",
|
| 57 |
"146": "NOT_ALIVE",
|
| 58 |
"147": "NON_PARTICIPATING",
|
| 59 |
"148": "NON_PARTICIPATING",
|
| 60 |
+
"149": "SUCCESS",
|
| 61 |
"15": "NOT_ALIVE",
|
| 62 |
"150": "SUCCESS",
|
| 63 |
"151": "NON_PARTICIPATING",
|
| 64 |
"152": "NON_PARTICIPATING",
|
| 65 |
"153": "NOT_ALIVE",
|
| 66 |
"154": "SUCCESS",
|
| 67 |
+
"155": "SUCCESS",
|
| 68 |
"156": "NOT_ALIVE",
|
| 69 |
+
"157": "SUCCESS",
|
| 70 |
"158": "NON_PARTICIPATING",
|
| 71 |
"159": "SUCCESS",
|
| 72 |
+
"16": "NOT_ALIVE",
|
| 73 |
"160": "NON_PARTICIPATING",
|
| 74 |
+
"161": "SUCCESS",
|
| 75 |
+
"162": "SUCCESS",
|
| 76 |
+
"163": "SUCCESS",
|
| 77 |
"164": "NON_PARTICIPATING",
|
| 78 |
+
"165": "SUCCESS",
|
| 79 |
+
"166": "NOT_ALIVE",
|
| 80 |
+
"167": "SUCCESS",
|
| 81 |
"168": "NON_PARTICIPATING",
|
| 82 |
"169": "NON_PARTICIPATING",
|
| 83 |
+
"17": "SUCCESS",
|
| 84 |
+
"170": "NON_PARTICIPATING",
|
| 85 |
+
"171": "NOT_ALIVE",
|
| 86 |
"172": "NOT_ALIVE",
|
| 87 |
"173": "NON_PARTICIPATING",
|
| 88 |
"174": "SUCCESS",
|
| 89 |
"175": "NON_PARTICIPATING",
|
| 90 |
"176": "NOT_ALIVE",
|
| 91 |
+
"177": "SUCCESS",
|
| 92 |
+
"178": "FAIL",
|
| 93 |
+
"179": "FAIL",
|
| 94 |
+
"18": "FAIL",
|
| 95 |
+
"180": "SUCCESS",
|
| 96 |
"181": "NON_PARTICIPATING",
|
| 97 |
+
"182": "NON_PARTICIPATING",
|
| 98 |
+
"183": "SUCCESS",
|
| 99 |
"184": "SUCCESS",
|
| 100 |
+
"185": "SUCCESS",
|
| 101 |
+
"186": "FAIL",
|
| 102 |
"187": "NON_PARTICIPATING",
|
| 103 |
"188": "SUCCESS",
|
| 104 |
+
"189": "NOT_ALIVE",
|
| 105 |
"19": "SUCCESS",
|
| 106 |
+
"190": "NON_PARTICIPATING",
|
| 107 |
+
"191": "FAIL",
|
| 108 |
"192": "SUCCESS",
|
| 109 |
"193": "SUCCESS",
|
| 110 |
"194": "NOT_ALIVE",
|
| 111 |
+
"195": "NON_PARTICIPATING",
|
| 112 |
+
"196": "SUCCESS",
|
| 113 |
"197": "NOT_ALIVE",
|
| 114 |
+
"198": "FAIL",
|
| 115 |
"199": "FAIL",
|
| 116 |
"2": "SUCCESS",
|
| 117 |
"20": "NOT_ALIVE",
|
| 118 |
"200": "SUCCESS",
|
| 119 |
"201": "NOT_ALIVE",
|
| 120 |
+
"202": "NON_PARTICIPATING",
|
| 121 |
+
"203": "FAIL",
|
| 122 |
"204": "NON_PARTICIPATING",
|
| 123 |
"205": "NON_PARTICIPATING",
|
| 124 |
"206": "SUCCESS",
|
| 125 |
"207": "NOT_ALIVE",
|
| 126 |
+
"208": "SUCCESS",
|
| 127 |
"209": "NON_PARTICIPATING",
|
| 128 |
"21": "NOT_ALIVE",
|
| 129 |
+
"210": "FAIL",
|
| 130 |
"211": "SUCCESS",
|
| 131 |
+
"212": "SUCCESS",
|
| 132 |
+
"213": "NON_PARTICIPATING",
|
| 133 |
+
"214": "NOT_ALIVE",
|
| 134 |
"215": "SUCCESS",
|
| 135 |
+
"216": "FAIL",
|
| 136 |
"217": "SUCCESS",
|
| 137 |
"218": "SUCCESS",
|
| 138 |
"219": "SUCCESS",
|
| 139 |
+
"22": "FAIL",
|
| 140 |
+
"220": "SUCCESS",
|
| 141 |
+
"221": "NON_PARTICIPATING",
|
| 142 |
+
"222": "SUCCESS",
|
| 143 |
"223": "NON_PARTICIPATING",
|
| 144 |
"224": "NON_PARTICIPATING",
|
| 145 |
"225": "NON_PARTICIPATING",
|
|
|
|
| 151 |
"230": "NOT_ALIVE",
|
| 152 |
"231": "NON_PARTICIPATING",
|
| 153 |
"232": "NOT_ALIVE",
|
| 154 |
+
"233": "NON_PARTICIPATING",
|
| 155 |
+
"234": "NOT_ALIVE",
|
| 156 |
"235": "NON_PARTICIPATING",
|
| 157 |
"236": "SUCCESS",
|
| 158 |
+
"237": "NON_PARTICIPATING",
|
| 159 |
"238": "NON_PARTICIPATING",
|
| 160 |
"239": "NOT_ALIVE",
|
| 161 |
+
"24": "NOT_ALIVE",
|
| 162 |
+
"240": "FAIL",
|
| 163 |
"241": "NON_PARTICIPATING",
|
| 164 |
+
"242": "NON_PARTICIPATING",
|
| 165 |
+
"243": "FAIL",
|
| 166 |
"244": "SUCCESS",
|
| 167 |
"245": "SUCCESS",
|
| 168 |
"246": "SUCCESS",
|
| 169 |
"247": "FAIL",
|
| 170 |
+
"248": "NOT_ALIVE",
|
| 171 |
"249": "SUCCESS",
|
| 172 |
"25": "SUCCESS",
|
| 173 |
"250": "FAIL",
|
|
|
|
| 175 |
"252": "SUCCESS",
|
| 176 |
"253": "SUCCESS",
|
| 177 |
"254": "NON_PARTICIPATING",
|
| 178 |
+
"255": "SUCCESS",
|
| 179 |
"26": "FAIL",
|
| 180 |
"27": "SUCCESS",
|
| 181 |
"28": "NON_PARTICIPATING",
|
|
|
|
| 185 |
"31": "SUCCESS",
|
| 186 |
"32": "SUCCESS",
|
| 187 |
"33": "NON_PARTICIPATING",
|
| 188 |
+
"34": "NON_PARTICIPATING",
|
| 189 |
"35": "FAIL",
|
| 190 |
+
"36": "FAIL",
|
| 191 |
+
"37": "NOT_ALIVE",
|
| 192 |
+
"38": "FAIL",
|
| 193 |
"39": "NOT_ALIVE",
|
| 194 |
"4": "NOT_ALIVE",
|
| 195 |
+
"40": "SUCCESS",
|
| 196 |
"41": "FAIL",
|
| 197 |
+
"42": "SUCCESS",
|
| 198 |
+
"43": "SUCCESS",
|
| 199 |
+
"44": "FAIL",
|
| 200 |
"45": "NON_PARTICIPATING",
|
| 201 |
"46": "NOT_ALIVE",
|
| 202 |
+
"47": "NON_PARTICIPATING",
|
| 203 |
+
"48": "FAIL",
|
| 204 |
"49": "NON_PARTICIPATING",
|
| 205 |
"5": "NOT_ALIVE",
|
| 206 |
"50": "NOT_ALIVE",
|
| 207 |
+
"51": "SUCCESS",
|
| 208 |
"52": "SUCCESS",
|
| 209 |
"53": "NOT_ALIVE",
|
| 210 |
+
"54": "SUCCESS",
|
| 211 |
"55": "SUCCESS",
|
| 212 |
"56": "FAIL",
|
| 213 |
"57": "SUCCESS",
|
| 214 |
"58": "SUCCESS",
|
| 215 |
+
"59": "NOT_ALIVE",
|
| 216 |
"6": "NOT_ALIVE",
|
| 217 |
+
"60": "NON_PARTICIPATING",
|
| 218 |
+
"61": "FAIL",
|
| 219 |
+
"62": "NOT_ALIVE",
|
| 220 |
+
"63": "NON_PARTICIPATING",
|
| 221 |
"64": "NON_PARTICIPATING",
|
| 222 |
"65": "SUCCESS",
|
| 223 |
"66": "NON_PARTICIPATING",
|
| 224 |
+
"67": "NON_PARTICIPATING",
|
| 225 |
+
"68": "NON_PARTICIPATING",
|
| 226 |
"69": "SUCCESS",
|
| 227 |
+
"7": "FAIL",
|
| 228 |
"70": "NOT_ALIVE",
|
| 229 |
"71": "SUCCESS",
|
| 230 |
"72": "NON_PARTICIPATING",
|
|
|
|
| 232 |
"74": "SUCCESS",
|
| 233 |
"75": "NOT_ALIVE",
|
| 234 |
"76": "SUCCESS",
|
| 235 |
+
"77": "NON_PARTICIPATING",
|
| 236 |
"78": "NON_PARTICIPATING",
|
| 237 |
"79": "SUCCESS",
|
| 238 |
"8": "SUCCESS",
|
| 239 |
"80": "NON_PARTICIPATING",
|
| 240 |
"81": "NON_PARTICIPATING",
|
| 241 |
+
"82": "SUCCESS",
|
| 242 |
"83": "SUCCESS",
|
| 243 |
"84": "SUCCESS",
|
| 244 |
+
"85": "SUCCESS",
|
| 245 |
"86": "NON_PARTICIPATING",
|
| 246 |
"87": "SUCCESS",
|
| 247 |
+
"88": "NOT_ALIVE",
|
| 248 |
+
"89": "SUCCESS",
|
| 249 |
"9": "NOT_ALIVE",
|
| 250 |
+
"90": "SUCCESS",
|
| 251 |
+
"91": "NOT_ALIVE",
|
| 252 |
"92": "SUCCESS",
|
| 253 |
+
"93": "NON_PARTICIPATING",
|
| 254 |
+
"94": "SUCCESS",
|
| 255 |
+
"95": "FAIL",
|
| 256 |
"96": "SUCCESS",
|
| 257 |
"97": "SUCCESS",
|
| 258 |
"98": "NOT_ALIVE",
|
|
|
|
| 264 |
"attention_bias": false,
|
| 265 |
"attention_dropout": 0.0,
|
| 266 |
"block_list": [
|
| 267 |
+
6402660,
|
| 268 |
+
6402677
|
| 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": 6402610,
|
| 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 sha256:8a5852d4281fb08079960dc17602b3bce4d8174226e2ac9c82c793e197f2ef63
|
| 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:b78cefff0ad5ca0c53c3d784842f4e53295c199ba3c703f64c4490f029ca9154
|
| 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 sha256:8f10f85fc4c7c6004052c65b6809ee189b241d969732a99f879d867fe460147b
|
| 3 |
size 4400216536
|