Run 8. Outer Step 46. Inner Step 4.
Browse files- config.json +76 -78
- gradients.pt +1 -1
- inner_optimizer.pt +1 -1
- model.safetensors +1 -1
config.json
CHANGED
|
@@ -10,14 +10,14 @@
|
|
| 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": "NON_PARTICIPATING",
|
| 19 |
"111": "NOT_ALIVE",
|
| 20 |
-
"112": "
|
| 21 |
"113": "NOT_ALIVE",
|
| 22 |
"114": "NOT_ALIVE",
|
| 23 |
"115": "NOT_ALIVE",
|
|
@@ -27,31 +27,31 @@
|
|
| 27 |
"119": "NOT_ALIVE",
|
| 28 |
"12": "NOT_ALIVE",
|
| 29 |
"120": "NOT_ALIVE",
|
| 30 |
-
"121": "
|
| 31 |
-
"122": "
|
| 32 |
"123": "SUCCESS",
|
| 33 |
"124": "NOT_ALIVE",
|
| 34 |
"125": "NOT_ALIVE",
|
| 35 |
-
"126": "
|
| 36 |
-
"127": "
|
| 37 |
"128": "NOT_ALIVE",
|
| 38 |
"129": "NOT_ALIVE",
|
| 39 |
"13": "NOT_ALIVE",
|
| 40 |
"130": "NON_PARTICIPATING",
|
| 41 |
-
"131": "
|
| 42 |
-
"132": "
|
| 43 |
-
"133": "
|
| 44 |
"134": "NON_PARTICIPATING",
|
| 45 |
-
"135": "
|
| 46 |
-
"136": "
|
| 47 |
-
"137": "
|
| 48 |
"138": "NOT_ALIVE",
|
| 49 |
"139": "NOT_ALIVE",
|
| 50 |
"14": "NOT_ALIVE",
|
| 51 |
"140": "SUCCESS",
|
| 52 |
"141": "NOT_ALIVE",
|
| 53 |
"142": "NON_PARTICIPATING",
|
| 54 |
-
"143": "
|
| 55 |
"144": "NOT_ALIVE",
|
| 56 |
"145": "SUCCESS",
|
| 57 |
"146": "NON_PARTICIPATING",
|
|
@@ -61,25 +61,25 @@
|
|
| 61 |
"15": "SUCCESS",
|
| 62 |
"150": "SUCCESS",
|
| 63 |
"151": "NON_PARTICIPATING",
|
| 64 |
-
"152": "
|
| 65 |
"153": "SUCCESS",
|
| 66 |
"154": "NOT_ALIVE",
|
| 67 |
-
"155": "
|
| 68 |
"156": "NOT_ALIVE",
|
| 69 |
"157": "NON_PARTICIPATING",
|
| 70 |
-
"158": "
|
| 71 |
"159": "NON_PARTICIPATING",
|
| 72 |
"16": "NOT_ALIVE",
|
| 73 |
-
"160": "
|
| 74 |
-
"161": "
|
| 75 |
"162": "NON_PARTICIPATING",
|
| 76 |
"163": "NOT_ALIVE",
|
| 77 |
"164": "FAIL",
|
| 78 |
"165": "SUCCESS",
|
| 79 |
"166": "NOT_ALIVE",
|
| 80 |
"167": "NOT_ALIVE",
|
| 81 |
-
"168": "
|
| 82 |
-
"169": "
|
| 83 |
"17": "NOT_ALIVE",
|
| 84 |
"170": "SUCCESS",
|
| 85 |
"171": "NON_PARTICIPATING",
|
|
@@ -88,38 +88,38 @@
|
|
| 88 |
"174": "SUCCESS",
|
| 89 |
"175": "NON_PARTICIPATING",
|
| 90 |
"176": "NOT_ALIVE",
|
| 91 |
-
"177": "
|
| 92 |
"178": "NON_PARTICIPATING",
|
| 93 |
-
"179": "
|
| 94 |
"18": "NOT_ALIVE",
|
| 95 |
-
"180": "
|
| 96 |
-
"181": "
|
| 97 |
"182": "NOT_ALIVE",
|
| 98 |
-
"183": "
|
| 99 |
"184": "NOT_ALIVE",
|
| 100 |
"185": "NON_PARTICIPATING",
|
| 101 |
-
"186": "
|
| 102 |
"187": "NOT_ALIVE",
|
| 103 |
-
"188": "
|
| 104 |
"189": "SUCCESS",
|
| 105 |
"19": "NOT_ALIVE",
|
| 106 |
"190": "NON_PARTICIPATING",
|
| 107 |
-
"191": "
|
| 108 |
-
"192": "
|
| 109 |
"193": "NOT_ALIVE",
|
| 110 |
"194": "NON_PARTICIPATING",
|
| 111 |
"195": "NON_PARTICIPATING",
|
| 112 |
-
"196": "
|
| 113 |
"197": "NOT_ALIVE",
|
| 114 |
"198": "NON_PARTICIPATING",
|
| 115 |
-
"199": "
|
| 116 |
"2": "NOT_ALIVE",
|
| 117 |
"20": "NOT_ALIVE",
|
| 118 |
"200": "NOT_ALIVE",
|
| 119 |
"201": "NOT_ALIVE",
|
| 120 |
"202": "FAIL",
|
| 121 |
"203": "FAIL",
|
| 122 |
-
"204": "
|
| 123 |
"205": "SUCCESS",
|
| 124 |
"206": "SUCCESS",
|
| 125 |
"207": "NOT_ALIVE",
|
|
@@ -131,24 +131,24 @@
|
|
| 131 |
"212": "SUCCESS",
|
| 132 |
"213": "NON_PARTICIPATING",
|
| 133 |
"214": "SUCCESS",
|
| 134 |
-
"215": "
|
| 135 |
-
"216": "
|
| 136 |
-
"217": "
|
| 137 |
"218": "FAIL",
|
| 138 |
-
"219": "
|
| 139 |
"22": "NOT_ALIVE",
|
| 140 |
-
"220": "
|
| 141 |
"221": "SUCCESS",
|
| 142 |
"222": "SUCCESS",
|
| 143 |
"223": "NOT_ALIVE",
|
| 144 |
"224": "NON_PARTICIPATING",
|
| 145 |
-
"225": "
|
| 146 |
"226": "NOT_ALIVE",
|
| 147 |
"227": "NON_PARTICIPATING",
|
| 148 |
"228": "SUCCESS",
|
| 149 |
-
"229": "
|
| 150 |
"23": "NOT_ALIVE",
|
| 151 |
-
"230": "
|
| 152 |
"231": "NON_PARTICIPATING",
|
| 153 |
"232": "NOT_ALIVE",
|
| 154 |
"233": "NON_PARTICIPATING",
|
|
@@ -158,9 +158,9 @@
|
|
| 158 |
"237": "NON_PARTICIPATING",
|
| 159 |
"238": "SUCCESS",
|
| 160 |
"239": "NON_PARTICIPATING",
|
| 161 |
-
"24": "
|
| 162 |
-
"240": "
|
| 163 |
-
"241": "
|
| 164 |
"242": "NON_PARTICIPATING",
|
| 165 |
"243": "NON_PARTICIPATING",
|
| 166 |
"244": "NON_PARTICIPATING",
|
|
@@ -172,38 +172,38 @@
|
|
| 172 |
"25": "SUCCESS",
|
| 173 |
"250": "NON_PARTICIPATING",
|
| 174 |
"251": "NOT_ALIVE",
|
| 175 |
-
"252": "
|
| 176 |
-
"253": "
|
| 177 |
-
"254": "
|
| 178 |
"255": "NON_PARTICIPATING",
|
| 179 |
-
"26": "
|
| 180 |
-
"27": "
|
| 181 |
"28": "NON_PARTICIPATING",
|
| 182 |
-
"29": "
|
| 183 |
-
"3": "
|
| 184 |
"30": "SUCCESS",
|
| 185 |
-
"31": "
|
| 186 |
-
"32": "
|
| 187 |
"33": "NOT_ALIVE",
|
| 188 |
"34": "NOT_ALIVE",
|
| 189 |
"35": "NON_PARTICIPATING",
|
| 190 |
-
"36": "
|
| 191 |
"37": "NOT_ALIVE",
|
| 192 |
"38": "SUCCESS",
|
| 193 |
"39": "NON_PARTICIPATING",
|
| 194 |
"4": "NOT_ALIVE",
|
| 195 |
"40": "NON_PARTICIPATING",
|
| 196 |
-
"41": "
|
| 197 |
"42": "NOT_ALIVE",
|
| 198 |
"43": "NON_PARTICIPATING",
|
| 199 |
-
"44": "
|
| 200 |
"45": "NOT_ALIVE",
|
| 201 |
"46": "NON_PARTICIPATING",
|
| 202 |
"47": "NOT_ALIVE",
|
| 203 |
"48": "NOT_ALIVE",
|
| 204 |
-
"49": "
|
| 205 |
"5": "NOT_ALIVE",
|
| 206 |
-
"50": "
|
| 207 |
"51": "NON_PARTICIPATING",
|
| 208 |
"52": "SUCCESS",
|
| 209 |
"53": "NON_PARTICIPATING",
|
|
@@ -214,33 +214,33 @@
|
|
| 214 |
"58": "NOT_ALIVE",
|
| 215 |
"59": "NOT_ALIVE",
|
| 216 |
"6": "NOT_ALIVE",
|
| 217 |
-
"60": "
|
| 218 |
-
"61": "
|
| 219 |
"62": "NOT_ALIVE",
|
| 220 |
-
"63": "
|
| 221 |
"64": "NOT_ALIVE",
|
| 222 |
"65": "NOT_ALIVE",
|
| 223 |
-
"66": "
|
| 224 |
"67": "NON_PARTICIPATING",
|
| 225 |
"68": "NON_PARTICIPATING",
|
| 226 |
-
"69": "
|
| 227 |
-
"7": "
|
| 228 |
"70": "NOT_ALIVE",
|
| 229 |
"71": "NON_PARTICIPATING",
|
| 230 |
"72": "NOT_ALIVE",
|
| 231 |
"73": "NOT_ALIVE",
|
| 232 |
-
"74": "
|
| 233 |
-
"75": "
|
| 234 |
"76": "NOT_ALIVE",
|
| 235 |
-
"77": "
|
| 236 |
"78": "NOT_ALIVE",
|
| 237 |
"79": "SUCCESS",
|
| 238 |
"8": "NOT_ALIVE",
|
| 239 |
"80": "NON_PARTICIPATING",
|
| 240 |
"81": "NON_PARTICIPATING",
|
| 241 |
-
"82": "
|
| 242 |
"83": "NON_PARTICIPATING",
|
| 243 |
-
"84": "
|
| 244 |
"85": "NOT_ALIVE",
|
| 245 |
"86": "NOT_ALIVE",
|
| 246 |
"87": "FAIL",
|
|
@@ -248,14 +248,14 @@
|
|
| 248 |
"89": "NON_PARTICIPATING",
|
| 249 |
"9": "NOT_ALIVE",
|
| 250 |
"90": "NOT_ALIVE",
|
| 251 |
-
"91": "
|
| 252 |
"92": "NOT_ALIVE",
|
| 253 |
"93": "NOT_ALIVE",
|
| 254 |
"94": "NOT_ALIVE",
|
| 255 |
"95": "NOT_ALIVE",
|
| 256 |
"96": "NON_PARTICIPATING",
|
| 257 |
-
"97": "
|
| 258 |
-
"98": "
|
| 259 |
"99": "NON_PARTICIPATING"
|
| 260 |
},
|
| 261 |
"architectures": [
|
|
@@ -264,19 +264,17 @@
|
|
| 264 |
"attention_bias": false,
|
| 265 |
"attention_dropout": 0.0,
|
| 266 |
"block_list": [
|
| 267 |
-
|
| 268 |
-
|
| 269 |
-
6371291,
|
| 270 |
-
6371307
|
| 271 |
],
|
| 272 |
"bos_token_id": 1,
|
| 273 |
"eos_token_id": 2,
|
| 274 |
"hidden_act": "silu",
|
| 275 |
"hidden_size": 2048,
|
| 276 |
"initializer_range": 0.02,
|
| 277 |
-
"inner_step":
|
| 278 |
"intermediate_size": 5632,
|
| 279 |
-
"last_allreduce_block":
|
| 280 |
"max_position_embeddings": 2048,
|
| 281 |
"mlp_bias": false,
|
| 282 |
"model_type": "llama",
|
|
|
|
| 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",
|
|
|
|
| 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": "NON_PARTICIPATING",
|
| 36 |
+
"127": "SUCCESS",
|
| 37 |
"128": "NOT_ALIVE",
|
| 38 |
"129": "NOT_ALIVE",
|
| 39 |
"13": "NOT_ALIVE",
|
| 40 |
"130": "NON_PARTICIPATING",
|
| 41 |
+
"131": "NOT_ALIVE",
|
| 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": "NOT_ALIVE",
|
| 50 |
"14": "NOT_ALIVE",
|
| 51 |
"140": "SUCCESS",
|
| 52 |
"141": "NOT_ALIVE",
|
| 53 |
"142": "NON_PARTICIPATING",
|
| 54 |
+
"143": "NOT_ALIVE",
|
| 55 |
"144": "NOT_ALIVE",
|
| 56 |
"145": "SUCCESS",
|
| 57 |
"146": "NON_PARTICIPATING",
|
|
|
|
| 61 |
"15": "SUCCESS",
|
| 62 |
"150": "SUCCESS",
|
| 63 |
"151": "NON_PARTICIPATING",
|
| 64 |
+
"152": "NON_PARTICIPATING",
|
| 65 |
"153": "SUCCESS",
|
| 66 |
"154": "NOT_ALIVE",
|
| 67 |
+
"155": "NON_PARTICIPATING",
|
| 68 |
"156": "NOT_ALIVE",
|
| 69 |
"157": "NON_PARTICIPATING",
|
| 70 |
+
"158": "NOT_ALIVE",
|
| 71 |
"159": "NON_PARTICIPATING",
|
| 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": "NON_PARTICIPATING",
|
| 82 |
+
"169": "NON_PARTICIPATING",
|
| 83 |
"17": "NOT_ALIVE",
|
| 84 |
"170": "SUCCESS",
|
| 85 |
"171": "NON_PARTICIPATING",
|
|
|
|
| 88 |
"174": "SUCCESS",
|
| 89 |
"175": "NON_PARTICIPATING",
|
| 90 |
"176": "NOT_ALIVE",
|
| 91 |
+
"177": "NON_PARTICIPATING",
|
| 92 |
"178": "NON_PARTICIPATING",
|
| 93 |
+
"179": "SUCCESS",
|
| 94 |
"18": "NOT_ALIVE",
|
| 95 |
+
"180": "NOT_ALIVE",
|
| 96 |
+
"181": "SUCCESS",
|
| 97 |
"182": "NOT_ALIVE",
|
| 98 |
+
"183": "NOT_ALIVE",
|
| 99 |
"184": "NOT_ALIVE",
|
| 100 |
"185": "NON_PARTICIPATING",
|
| 101 |
+
"186": "NOT_ALIVE",
|
| 102 |
"187": "NOT_ALIVE",
|
| 103 |
+
"188": "SUCCESS",
|
| 104 |
"189": "SUCCESS",
|
| 105 |
"19": "NOT_ALIVE",
|
| 106 |
"190": "NON_PARTICIPATING",
|
| 107 |
+
"191": "FAIL",
|
| 108 |
+
"192": "SUCCESS",
|
| 109 |
"193": "NOT_ALIVE",
|
| 110 |
"194": "NON_PARTICIPATING",
|
| 111 |
"195": "NON_PARTICIPATING",
|
| 112 |
+
"196": "SUCCESS",
|
| 113 |
"197": "NOT_ALIVE",
|
| 114 |
"198": "NON_PARTICIPATING",
|
| 115 |
+
"199": "SUCCESS",
|
| 116 |
"2": "NOT_ALIVE",
|
| 117 |
"20": "NOT_ALIVE",
|
| 118 |
"200": "NOT_ALIVE",
|
| 119 |
"201": "NOT_ALIVE",
|
| 120 |
"202": "FAIL",
|
| 121 |
"203": "FAIL",
|
| 122 |
+
"204": "NON_PARTICIPATING",
|
| 123 |
"205": "SUCCESS",
|
| 124 |
"206": "SUCCESS",
|
| 125 |
"207": "NOT_ALIVE",
|
|
|
|
| 131 |
"212": "SUCCESS",
|
| 132 |
"213": "NON_PARTICIPATING",
|
| 133 |
"214": "SUCCESS",
|
| 134 |
+
"215": "NOT_ALIVE",
|
| 135 |
+
"216": "SUCCESS",
|
| 136 |
+
"217": "SUCCESS",
|
| 137 |
"218": "FAIL",
|
| 138 |
+
"219": "SUCCESS",
|
| 139 |
"22": "NOT_ALIVE",
|
| 140 |
+
"220": "NOT_ALIVE",
|
| 141 |
"221": "SUCCESS",
|
| 142 |
"222": "SUCCESS",
|
| 143 |
"223": "NOT_ALIVE",
|
| 144 |
"224": "NON_PARTICIPATING",
|
| 145 |
+
"225": "SUCCESS",
|
| 146 |
"226": "NOT_ALIVE",
|
| 147 |
"227": "NON_PARTICIPATING",
|
| 148 |
"228": "SUCCESS",
|
| 149 |
+
"229": "NON_PARTICIPATING",
|
| 150 |
"23": "NOT_ALIVE",
|
| 151 |
+
"230": "FAIL",
|
| 152 |
"231": "NON_PARTICIPATING",
|
| 153 |
"232": "NOT_ALIVE",
|
| 154 |
"233": "NON_PARTICIPATING",
|
|
|
|
| 158 |
"237": "NON_PARTICIPATING",
|
| 159 |
"238": "SUCCESS",
|
| 160 |
"239": "NON_PARTICIPATING",
|
| 161 |
+
"24": "NOT_ALIVE",
|
| 162 |
+
"240": "NOT_ALIVE",
|
| 163 |
+
"241": "NON_PARTICIPATING",
|
| 164 |
"242": "NON_PARTICIPATING",
|
| 165 |
"243": "NON_PARTICIPATING",
|
| 166 |
"244": "NON_PARTICIPATING",
|
|
|
|
| 172 |
"25": "SUCCESS",
|
| 173 |
"250": "NON_PARTICIPATING",
|
| 174 |
"251": "NOT_ALIVE",
|
| 175 |
+
"252": "SUCCESS",
|
| 176 |
+
"253": "NON_PARTICIPATING",
|
| 177 |
+
"254": "NOT_ALIVE",
|
| 178 |
"255": "NON_PARTICIPATING",
|
| 179 |
+
"26": "SUCCESS",
|
| 180 |
+
"27": "SUCCESS",
|
| 181 |
"28": "NON_PARTICIPATING",
|
| 182 |
+
"29": "SUCCESS",
|
| 183 |
+
"3": "SUCCESS",
|
| 184 |
"30": "SUCCESS",
|
| 185 |
+
"31": "NON_PARTICIPATING",
|
| 186 |
+
"32": "NOT_ALIVE",
|
| 187 |
"33": "NOT_ALIVE",
|
| 188 |
"34": "NOT_ALIVE",
|
| 189 |
"35": "NON_PARTICIPATING",
|
| 190 |
+
"36": "NOT_ALIVE",
|
| 191 |
"37": "NOT_ALIVE",
|
| 192 |
"38": "SUCCESS",
|
| 193 |
"39": "NON_PARTICIPATING",
|
| 194 |
"4": "NOT_ALIVE",
|
| 195 |
"40": "NON_PARTICIPATING",
|
| 196 |
+
"41": "NOT_ALIVE",
|
| 197 |
"42": "NOT_ALIVE",
|
| 198 |
"43": "NON_PARTICIPATING",
|
| 199 |
+
"44": "FAIL",
|
| 200 |
"45": "NOT_ALIVE",
|
| 201 |
"46": "NON_PARTICIPATING",
|
| 202 |
"47": "NOT_ALIVE",
|
| 203 |
"48": "NOT_ALIVE",
|
| 204 |
+
"49": "NOT_ALIVE",
|
| 205 |
"5": "NOT_ALIVE",
|
| 206 |
+
"50": "SUCCESS",
|
| 207 |
"51": "NON_PARTICIPATING",
|
| 208 |
"52": "SUCCESS",
|
| 209 |
"53": "NON_PARTICIPATING",
|
|
|
|
| 214 |
"58": "NOT_ALIVE",
|
| 215 |
"59": "NOT_ALIVE",
|
| 216 |
"6": "NOT_ALIVE",
|
| 217 |
+
"60": "NON_PARTICIPATING",
|
| 218 |
+
"61": "NOT_ALIVE",
|
| 219 |
"62": "NOT_ALIVE",
|
| 220 |
+
"63": "SUCCESS",
|
| 221 |
"64": "NOT_ALIVE",
|
| 222 |
"65": "NOT_ALIVE",
|
| 223 |
+
"66": "NOT_ALIVE",
|
| 224 |
"67": "NON_PARTICIPATING",
|
| 225 |
"68": "NON_PARTICIPATING",
|
| 226 |
+
"69": "NON_PARTICIPATING",
|
| 227 |
+
"7": "NOT_ALIVE",
|
| 228 |
"70": "NOT_ALIVE",
|
| 229 |
"71": "NON_PARTICIPATING",
|
| 230 |
"72": "NOT_ALIVE",
|
| 231 |
"73": "NOT_ALIVE",
|
| 232 |
+
"74": "FAIL",
|
| 233 |
+
"75": "NON_PARTICIPATING",
|
| 234 |
"76": "NOT_ALIVE",
|
| 235 |
+
"77": "SUCCESS",
|
| 236 |
"78": "NOT_ALIVE",
|
| 237 |
"79": "SUCCESS",
|
| 238 |
"8": "NOT_ALIVE",
|
| 239 |
"80": "NON_PARTICIPATING",
|
| 240 |
"81": "NON_PARTICIPATING",
|
| 241 |
+
"82": "NOT_ALIVE",
|
| 242 |
"83": "NON_PARTICIPATING",
|
| 243 |
+
"84": "NON_PARTICIPATING",
|
| 244 |
"85": "NOT_ALIVE",
|
| 245 |
"86": "NOT_ALIVE",
|
| 246 |
"87": "FAIL",
|
|
|
|
| 248 |
"89": "NON_PARTICIPATING",
|
| 249 |
"9": "NOT_ALIVE",
|
| 250 |
"90": "NOT_ALIVE",
|
| 251 |
+
"91": "NON_PARTICIPATING",
|
| 252 |
"92": "NOT_ALIVE",
|
| 253 |
"93": "NOT_ALIVE",
|
| 254 |
"94": "NOT_ALIVE",
|
| 255 |
"95": "NOT_ALIVE",
|
| 256 |
"96": "NON_PARTICIPATING",
|
| 257 |
+
"97": "SUCCESS",
|
| 258 |
+
"98": "SUCCESS",
|
| 259 |
"99": "NON_PARTICIPATING"
|
| 260 |
},
|
| 261 |
"architectures": [
|
|
|
|
| 264 |
"attention_bias": false,
|
| 265 |
"attention_dropout": 0.0,
|
| 266 |
"block_list": [
|
| 267 |
+
6371463,
|
| 268 |
+
6371479
|
|
|
|
|
|
|
| 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": 4,
|
| 276 |
"intermediate_size": 5632,
|
| 277 |
+
"last_allreduce_block": 6371430,
|
| 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:ab8cc5de61104a71af5aa059715cf8392f2150159a88327135e94dcf5b0cd491
|
| 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 sha256:279f9fce578d6d97c511058f1d3ecb67d8c11dd01536eea7fe346564db415934
|
| 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:f48fa9e1e5e21bf32ab1cf555e63cd645c38d2540006f2b15505b0705e5ee381
|
| 3 |
size 4400216536
|