Run 8. Outer Step 132. Inner Step 4.
Browse files- config.json +65 -65
- gradients.pt +1 -1
- inner_optimizer.pt +1 -1
- model.safetensors +1 -1
config.json
CHANGED
|
@@ -9,7 +9,7 @@
|
|
| 9 |
"102": "NOT_ALIVE",
|
| 10 |
"103": "NON_PARTICIPATING",
|
| 11 |
"104": "NON_PARTICIPATING",
|
| 12 |
-
"105": "
|
| 13 |
"106": "NOT_ALIVE",
|
| 14 |
"107": "NON_PARTICIPATING",
|
| 15 |
"108": "NON_PARTICIPATING",
|
|
@@ -17,19 +17,19 @@
|
|
| 17 |
"11": "NOT_ALIVE",
|
| 18 |
"110": "NON_PARTICIPATING",
|
| 19 |
"111": "NON_PARTICIPATING",
|
| 20 |
-
"112": "
|
| 21 |
"113": "NON_PARTICIPATING",
|
| 22 |
-
"114": "
|
| 23 |
"115": "NON_PARTICIPATING",
|
| 24 |
"116": "NON_PARTICIPATING",
|
| 25 |
-
"117": "
|
| 26 |
"118": "NON_PARTICIPATING",
|
| 27 |
"119": "NON_PARTICIPATING",
|
| 28 |
"12": "NOT_ALIVE",
|
| 29 |
"120": "NOT_ALIVE",
|
| 30 |
-
"121": "
|
| 31 |
"122": "NON_PARTICIPATING",
|
| 32 |
-
"123": "
|
| 33 |
"124": "NON_PARTICIPATING",
|
| 34 |
"125": "NOT_ALIVE",
|
| 35 |
"126": "NOT_ALIVE",
|
|
@@ -39,14 +39,14 @@
|
|
| 39 |
"13": "SUCCESS",
|
| 40 |
"130": "NON_PARTICIPATING",
|
| 41 |
"131": "NON_PARTICIPATING",
|
| 42 |
-
"132": "
|
| 43 |
-
"133": "
|
| 44 |
"134": "NON_PARTICIPATING",
|
| 45 |
"135": "NOT_ALIVE",
|
| 46 |
-
"136": "
|
| 47 |
-
"137": "
|
| 48 |
-
"138": "
|
| 49 |
-
"139": "
|
| 50 |
"14": "NOT_ALIVE",
|
| 51 |
"140": "NON_PARTICIPATING",
|
| 52 |
"141": "SUCCESS",
|
|
@@ -57,65 +57,65 @@
|
|
| 57 |
"146": "SUCCESS",
|
| 58 |
"147": "NOT_ALIVE",
|
| 59 |
"148": "NON_PARTICIPATING",
|
| 60 |
-
"149": "
|
| 61 |
"15": "NOT_ALIVE",
|
| 62 |
"150": "NON_PARTICIPATING",
|
| 63 |
"151": "NON_PARTICIPATING",
|
| 64 |
-
"152": "
|
| 65 |
-
"153": "
|
| 66 |
"154": "NOT_ALIVE",
|
| 67 |
"155": "NON_PARTICIPATING",
|
| 68 |
"156": "NOT_ALIVE",
|
| 69 |
"157": "NOT_ALIVE",
|
| 70 |
"158": "SUCCESS",
|
| 71 |
-
"159": "
|
| 72 |
"16": "NOT_ALIVE",
|
| 73 |
"160": "NON_PARTICIPATING",
|
| 74 |
"161": "NOT_ALIVE",
|
| 75 |
"162": "NON_PARTICIPATING",
|
| 76 |
"163": "SUCCESS",
|
| 77 |
-
"164": "
|
| 78 |
"165": "SUCCESS",
|
| 79 |
-
"166": "
|
| 80 |
"167": "NON_PARTICIPATING",
|
| 81 |
"168": "SUCCESS",
|
| 82 |
"169": "NON_PARTICIPATING",
|
| 83 |
"17": "NOT_ALIVE",
|
| 84 |
"170": "NOT_ALIVE",
|
| 85 |
-
"171": "
|
| 86 |
"172": "NON_PARTICIPATING",
|
| 87 |
"173": "SUCCESS",
|
| 88 |
-
"174": "
|
| 89 |
"175": "SUCCESS",
|
| 90 |
"176": "NOT_ALIVE",
|
| 91 |
-
"177": "
|
| 92 |
"178": "SUCCESS",
|
| 93 |
-
"179": "
|
| 94 |
"18": "NOT_ALIVE",
|
| 95 |
"180": "NOT_ALIVE",
|
| 96 |
"181": "NON_PARTICIPATING",
|
| 97 |
"182": "SUCCESS",
|
| 98 |
"183": "SUCCESS",
|
| 99 |
"184": "SUCCESS",
|
| 100 |
-
"185": "
|
| 101 |
"186": "NOT_ALIVE",
|
| 102 |
-
"187": "
|
| 103 |
"188": "NON_PARTICIPATING",
|
| 104 |
"189": "SUCCESS",
|
| 105 |
"19": "NOT_ALIVE",
|
| 106 |
-
"190": "
|
| 107 |
"191": "NON_PARTICIPATING",
|
| 108 |
-
"192": "
|
| 109 |
"193": "NON_PARTICIPATING",
|
| 110 |
-
"194": "
|
| 111 |
-
"195": "
|
| 112 |
-
"196": "
|
| 113 |
"197": "NOT_ALIVE",
|
| 114 |
"198": "SUCCESS",
|
| 115 |
"199": "SUCCESS",
|
| 116 |
"2": "NOT_ALIVE",
|
| 117 |
"20": "NOT_ALIVE",
|
| 118 |
-
"200": "
|
| 119 |
"201": "NOT_ALIVE",
|
| 120 |
"202": "SUCCESS",
|
| 121 |
"203": "SUCCESS",
|
|
@@ -126,11 +126,11 @@
|
|
| 126 |
"208": "NON_PARTICIPATING",
|
| 127 |
"209": "NON_PARTICIPATING",
|
| 128 |
"21": "NOT_ALIVE",
|
| 129 |
-
"210": "
|
| 130 |
"211": "NON_PARTICIPATING",
|
| 131 |
"212": "NON_PARTICIPATING",
|
| 132 |
-
"213": "
|
| 133 |
-
"214": "
|
| 134 |
"215": "NON_PARTICIPATING",
|
| 135 |
"216": "SUCCESS",
|
| 136 |
"217": "SUCCESS",
|
|
@@ -142,21 +142,21 @@
|
|
| 142 |
"222": "NON_PARTICIPATING",
|
| 143 |
"223": "NON_PARTICIPATING",
|
| 144 |
"224": "NON_PARTICIPATING",
|
| 145 |
-
"225": "
|
| 146 |
"226": "NON_PARTICIPATING",
|
| 147 |
"227": "SUCCESS",
|
| 148 |
"228": "SUCCESS",
|
| 149 |
"229": "NOT_ALIVE",
|
| 150 |
"23": "NOT_ALIVE",
|
| 151 |
-
"230": "
|
| 152 |
"231": "SUCCESS",
|
| 153 |
"232": "NOT_ALIVE",
|
| 154 |
-
"233": "
|
| 155 |
-
"234": "
|
| 156 |
-
"235": "
|
| 157 |
"236": "NON_PARTICIPATING",
|
| 158 |
"237": "SUCCESS",
|
| 159 |
-
"238": "
|
| 160 |
"239": "NON_PARTICIPATING",
|
| 161 |
"24": "NOT_ALIVE",
|
| 162 |
"240": "NON_PARTICIPATING",
|
|
@@ -165,30 +165,30 @@
|
|
| 165 |
"243": "NON_PARTICIPATING",
|
| 166 |
"244": "NON_PARTICIPATING",
|
| 167 |
"245": "NON_PARTICIPATING",
|
| 168 |
-
"246": "
|
| 169 |
-
"247": "
|
| 170 |
-
"248": "
|
| 171 |
-
"249": "
|
| 172 |
"25": "SUCCESS",
|
| 173 |
-
"250": "
|
| 174 |
-
"251": "
|
| 175 |
"252": "NON_PARTICIPATING",
|
| 176 |
"253": "SUCCESS",
|
| 177 |
"254": "NOT_ALIVE",
|
| 178 |
"255": "NOT_ALIVE",
|
| 179 |
"26": "NON_PARTICIPATING",
|
| 180 |
-
"27": "
|
| 181 |
"28": "NON_PARTICIPATING",
|
| 182 |
"29": "NON_PARTICIPATING",
|
| 183 |
"3": "NOT_ALIVE",
|
| 184 |
-
"30": "
|
| 185 |
"31": "NON_PARTICIPATING",
|
| 186 |
"32": "NON_PARTICIPATING",
|
| 187 |
"33": "NON_PARTICIPATING",
|
| 188 |
"34": "NON_PARTICIPATING",
|
| 189 |
"35": "NON_PARTICIPATING",
|
| 190 |
-
"36": "
|
| 191 |
-
"37": "
|
| 192 |
"38": "SUCCESS",
|
| 193 |
"39": "NOT_ALIVE",
|
| 194 |
"4": "NOT_ALIVE",
|
|
@@ -203,14 +203,14 @@
|
|
| 203 |
"48": "NON_PARTICIPATING",
|
| 204 |
"49": "SUCCESS",
|
| 205 |
"5": "NOT_ALIVE",
|
| 206 |
-
"50": "
|
| 207 |
"51": "NON_PARTICIPATING",
|
| 208 |
"52": "SUCCESS",
|
| 209 |
"53": "NON_PARTICIPATING",
|
| 210 |
"54": "NON_PARTICIPATING",
|
| 211 |
-
"55": "
|
| 212 |
"56": "NON_PARTICIPATING",
|
| 213 |
-
"57": "
|
| 214 |
"58": "NOT_ALIVE",
|
| 215 |
"59": "NOT_ALIVE",
|
| 216 |
"6": "NOT_ALIVE",
|
|
@@ -218,18 +218,18 @@
|
|
| 218 |
"61": "SUCCESS",
|
| 219 |
"62": "NOT_ALIVE",
|
| 220 |
"63": "NON_PARTICIPATING",
|
| 221 |
-
"64": "
|
| 222 |
"65": "NON_PARTICIPATING",
|
| 223 |
"66": "SUCCESS",
|
| 224 |
-
"67": "
|
| 225 |
-
"68": "
|
| 226 |
"69": "SUCCESS",
|
| 227 |
"7": "NOT_ALIVE",
|
| 228 |
"70": "NON_PARTICIPATING",
|
| 229 |
"71": "NON_PARTICIPATING",
|
| 230 |
"72": "NON_PARTICIPATING",
|
| 231 |
"73": "NON_PARTICIPATING",
|
| 232 |
-
"74": "
|
| 233 |
"75": "NON_PARTICIPATING",
|
| 234 |
"76": "NON_PARTICIPATING",
|
| 235 |
"77": "NON_PARTICIPATING",
|
|
@@ -238,12 +238,12 @@
|
|
| 238 |
"8": "NOT_ALIVE",
|
| 239 |
"80": "NOT_ALIVE",
|
| 240 |
"81": "NON_PARTICIPATING",
|
| 241 |
-
"82": "
|
| 242 |
"83": "NOT_ALIVE",
|
| 243 |
"84": "NON_PARTICIPATING",
|
| 244 |
-
"85": "
|
| 245 |
"86": "NON_PARTICIPATING",
|
| 246 |
-
"87": "
|
| 247 |
"88": "SUCCESS",
|
| 248 |
"89": "NON_PARTICIPATING",
|
| 249 |
"9": "NOT_ALIVE",
|
|
@@ -253,7 +253,7 @@
|
|
| 253 |
"93": "SUCCESS",
|
| 254 |
"94": "SUCCESS",
|
| 255 |
"95": "NON_PARTICIPATING",
|
| 256 |
-
"96": "
|
| 257 |
"97": "NON_PARTICIPATING",
|
| 258 |
"98": "SUCCESS",
|
| 259 |
"99": "SUCCESS"
|
|
@@ -264,18 +264,18 @@
|
|
| 264 |
"attention_bias": false,
|
| 265 |
"attention_dropout": 0.0,
|
| 266 |
"block_list": [
|
| 267 |
-
|
| 268 |
-
|
| 269 |
-
|
| 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",
|
|
|
|
| 9 |
"102": "NOT_ALIVE",
|
| 10 |
"103": "NON_PARTICIPATING",
|
| 11 |
"104": "NON_PARTICIPATING",
|
| 12 |
+
"105": "NON_PARTICIPATING",
|
| 13 |
"106": "NOT_ALIVE",
|
| 14 |
"107": "NON_PARTICIPATING",
|
| 15 |
"108": "NON_PARTICIPATING",
|
|
|
|
| 17 |
"11": "NOT_ALIVE",
|
| 18 |
"110": "NON_PARTICIPATING",
|
| 19 |
"111": "NON_PARTICIPATING",
|
| 20 |
+
"112": "SUCCESS",
|
| 21 |
"113": "NON_PARTICIPATING",
|
| 22 |
+
"114": "NON_PARTICIPATING",
|
| 23 |
"115": "NON_PARTICIPATING",
|
| 24 |
"116": "NON_PARTICIPATING",
|
| 25 |
+
"117": "SUCCESS",
|
| 26 |
"118": "NON_PARTICIPATING",
|
| 27 |
"119": "NON_PARTICIPATING",
|
| 28 |
"12": "NOT_ALIVE",
|
| 29 |
"120": "NOT_ALIVE",
|
| 30 |
+
"121": "NOT_ALIVE",
|
| 31 |
"122": "NON_PARTICIPATING",
|
| 32 |
+
"123": "SUCCESS",
|
| 33 |
"124": "NON_PARTICIPATING",
|
| 34 |
"125": "NOT_ALIVE",
|
| 35 |
"126": "NOT_ALIVE",
|
|
|
|
| 39 |
"13": "SUCCESS",
|
| 40 |
"130": "NON_PARTICIPATING",
|
| 41 |
"131": "NON_PARTICIPATING",
|
| 42 |
+
"132": "NON_PARTICIPATING",
|
| 43 |
+
"133": "NON_PARTICIPATING",
|
| 44 |
"134": "NON_PARTICIPATING",
|
| 45 |
"135": "NOT_ALIVE",
|
| 46 |
+
"136": "NOT_ALIVE",
|
| 47 |
+
"137": "NON_PARTICIPATING",
|
| 48 |
+
"138": "NON_PARTICIPATING",
|
| 49 |
+
"139": "NON_PARTICIPATING",
|
| 50 |
"14": "NOT_ALIVE",
|
| 51 |
"140": "NON_PARTICIPATING",
|
| 52 |
"141": "SUCCESS",
|
|
|
|
| 57 |
"146": "SUCCESS",
|
| 58 |
"147": "NOT_ALIVE",
|
| 59 |
"148": "NON_PARTICIPATING",
|
| 60 |
+
"149": "NOT_ALIVE",
|
| 61 |
"15": "NOT_ALIVE",
|
| 62 |
"150": "NON_PARTICIPATING",
|
| 63 |
"151": "NON_PARTICIPATING",
|
| 64 |
+
"152": "SUCCESS",
|
| 65 |
+
"153": "NON_PARTICIPATING",
|
| 66 |
"154": "NOT_ALIVE",
|
| 67 |
"155": "NON_PARTICIPATING",
|
| 68 |
"156": "NOT_ALIVE",
|
| 69 |
"157": "NOT_ALIVE",
|
| 70 |
"158": "SUCCESS",
|
| 71 |
+
"159": "SUCCESS",
|
| 72 |
"16": "NOT_ALIVE",
|
| 73 |
"160": "NON_PARTICIPATING",
|
| 74 |
"161": "NOT_ALIVE",
|
| 75 |
"162": "NON_PARTICIPATING",
|
| 76 |
"163": "SUCCESS",
|
| 77 |
+
"164": "NOT_ALIVE",
|
| 78 |
"165": "SUCCESS",
|
| 79 |
+
"166": "SUCCESS",
|
| 80 |
"167": "NON_PARTICIPATING",
|
| 81 |
"168": "SUCCESS",
|
| 82 |
"169": "NON_PARTICIPATING",
|
| 83 |
"17": "NOT_ALIVE",
|
| 84 |
"170": "NOT_ALIVE",
|
| 85 |
+
"171": "NON_PARTICIPATING",
|
| 86 |
"172": "NON_PARTICIPATING",
|
| 87 |
"173": "SUCCESS",
|
| 88 |
+
"174": "NOT_ALIVE",
|
| 89 |
"175": "SUCCESS",
|
| 90 |
"176": "NOT_ALIVE",
|
| 91 |
+
"177": "NON_PARTICIPATING",
|
| 92 |
"178": "SUCCESS",
|
| 93 |
+
"179": "NOT_ALIVE",
|
| 94 |
"18": "NOT_ALIVE",
|
| 95 |
"180": "NOT_ALIVE",
|
| 96 |
"181": "NON_PARTICIPATING",
|
| 97 |
"182": "SUCCESS",
|
| 98 |
"183": "SUCCESS",
|
| 99 |
"184": "SUCCESS",
|
| 100 |
+
"185": "NON_PARTICIPATING",
|
| 101 |
"186": "NOT_ALIVE",
|
| 102 |
+
"187": "NOT_ALIVE",
|
| 103 |
"188": "NON_PARTICIPATING",
|
| 104 |
"189": "SUCCESS",
|
| 105 |
"19": "NOT_ALIVE",
|
| 106 |
+
"190": "NON_PARTICIPATING",
|
| 107 |
"191": "NON_PARTICIPATING",
|
| 108 |
+
"192": "NOT_ALIVE",
|
| 109 |
"193": "NON_PARTICIPATING",
|
| 110 |
+
"194": "NON_PARTICIPATING",
|
| 111 |
+
"195": "SUCCESS",
|
| 112 |
+
"196": "SUCCESS",
|
| 113 |
"197": "NOT_ALIVE",
|
| 114 |
"198": "SUCCESS",
|
| 115 |
"199": "SUCCESS",
|
| 116 |
"2": "NOT_ALIVE",
|
| 117 |
"20": "NOT_ALIVE",
|
| 118 |
+
"200": "SUCCESS",
|
| 119 |
"201": "NOT_ALIVE",
|
| 120 |
"202": "SUCCESS",
|
| 121 |
"203": "SUCCESS",
|
|
|
|
| 126 |
"208": "NON_PARTICIPATING",
|
| 127 |
"209": "NON_PARTICIPATING",
|
| 128 |
"21": "NOT_ALIVE",
|
| 129 |
+
"210": "SUCCESS",
|
| 130 |
"211": "NON_PARTICIPATING",
|
| 131 |
"212": "NON_PARTICIPATING",
|
| 132 |
+
"213": "NOT_ALIVE",
|
| 133 |
+
"214": "SUCCESS",
|
| 134 |
"215": "NON_PARTICIPATING",
|
| 135 |
"216": "SUCCESS",
|
| 136 |
"217": "SUCCESS",
|
|
|
|
| 142 |
"222": "NON_PARTICIPATING",
|
| 143 |
"223": "NON_PARTICIPATING",
|
| 144 |
"224": "NON_PARTICIPATING",
|
| 145 |
+
"225": "NON_PARTICIPATING",
|
| 146 |
"226": "NON_PARTICIPATING",
|
| 147 |
"227": "SUCCESS",
|
| 148 |
"228": "SUCCESS",
|
| 149 |
"229": "NOT_ALIVE",
|
| 150 |
"23": "NOT_ALIVE",
|
| 151 |
+
"230": "NON_PARTICIPATING",
|
| 152 |
"231": "SUCCESS",
|
| 153 |
"232": "NOT_ALIVE",
|
| 154 |
+
"233": "NOT_ALIVE",
|
| 155 |
+
"234": "NON_PARTICIPATING",
|
| 156 |
+
"235": "NON_PARTICIPATING",
|
| 157 |
"236": "NON_PARTICIPATING",
|
| 158 |
"237": "SUCCESS",
|
| 159 |
+
"238": "NOT_ALIVE",
|
| 160 |
"239": "NON_PARTICIPATING",
|
| 161 |
"24": "NOT_ALIVE",
|
| 162 |
"240": "NON_PARTICIPATING",
|
|
|
|
| 165 |
"243": "NON_PARTICIPATING",
|
| 166 |
"244": "NON_PARTICIPATING",
|
| 167 |
"245": "NON_PARTICIPATING",
|
| 168 |
+
"246": "SUCCESS",
|
| 169 |
+
"247": "FAIL",
|
| 170 |
+
"248": "NON_PARTICIPATING",
|
| 171 |
+
"249": "NOT_ALIVE",
|
| 172 |
"25": "SUCCESS",
|
| 173 |
+
"250": "NOT_ALIVE",
|
| 174 |
+
"251": "NOT_ALIVE",
|
| 175 |
"252": "NON_PARTICIPATING",
|
| 176 |
"253": "SUCCESS",
|
| 177 |
"254": "NOT_ALIVE",
|
| 178 |
"255": "NOT_ALIVE",
|
| 179 |
"26": "NON_PARTICIPATING",
|
| 180 |
+
"27": "NON_PARTICIPATING",
|
| 181 |
"28": "NON_PARTICIPATING",
|
| 182 |
"29": "NON_PARTICIPATING",
|
| 183 |
"3": "NOT_ALIVE",
|
| 184 |
+
"30": "NOT_ALIVE",
|
| 185 |
"31": "NON_PARTICIPATING",
|
| 186 |
"32": "NON_PARTICIPATING",
|
| 187 |
"33": "NON_PARTICIPATING",
|
| 188 |
"34": "NON_PARTICIPATING",
|
| 189 |
"35": "NON_PARTICIPATING",
|
| 190 |
+
"36": "NON_PARTICIPATING",
|
| 191 |
+
"37": "NON_PARTICIPATING",
|
| 192 |
"38": "SUCCESS",
|
| 193 |
"39": "NOT_ALIVE",
|
| 194 |
"4": "NOT_ALIVE",
|
|
|
|
| 203 |
"48": "NON_PARTICIPATING",
|
| 204 |
"49": "SUCCESS",
|
| 205 |
"5": "NOT_ALIVE",
|
| 206 |
+
"50": "SUCCESS",
|
| 207 |
"51": "NON_PARTICIPATING",
|
| 208 |
"52": "SUCCESS",
|
| 209 |
"53": "NON_PARTICIPATING",
|
| 210 |
"54": "NON_PARTICIPATING",
|
| 211 |
+
"55": "NON_PARTICIPATING",
|
| 212 |
"56": "NON_PARTICIPATING",
|
| 213 |
+
"57": "NOT_ALIVE",
|
| 214 |
"58": "NOT_ALIVE",
|
| 215 |
"59": "NOT_ALIVE",
|
| 216 |
"6": "NOT_ALIVE",
|
|
|
|
| 218 |
"61": "SUCCESS",
|
| 219 |
"62": "NOT_ALIVE",
|
| 220 |
"63": "NON_PARTICIPATING",
|
| 221 |
+
"64": "NON_PARTICIPATING",
|
| 222 |
"65": "NON_PARTICIPATING",
|
| 223 |
"66": "SUCCESS",
|
| 224 |
+
"67": "NOT_ALIVE",
|
| 225 |
+
"68": "NON_PARTICIPATING",
|
| 226 |
"69": "SUCCESS",
|
| 227 |
"7": "NOT_ALIVE",
|
| 228 |
"70": "NON_PARTICIPATING",
|
| 229 |
"71": "NON_PARTICIPATING",
|
| 230 |
"72": "NON_PARTICIPATING",
|
| 231 |
"73": "NON_PARTICIPATING",
|
| 232 |
+
"74": "NON_PARTICIPATING",
|
| 233 |
"75": "NON_PARTICIPATING",
|
| 234 |
"76": "NON_PARTICIPATING",
|
| 235 |
"77": "NON_PARTICIPATING",
|
|
|
|
| 238 |
"8": "NOT_ALIVE",
|
| 239 |
"80": "NOT_ALIVE",
|
| 240 |
"81": "NON_PARTICIPATING",
|
| 241 |
+
"82": "NOT_ALIVE",
|
| 242 |
"83": "NOT_ALIVE",
|
| 243 |
"84": "NON_PARTICIPATING",
|
| 244 |
+
"85": "NON_PARTICIPATING",
|
| 245 |
"86": "NON_PARTICIPATING",
|
| 246 |
+
"87": "FAIL",
|
| 247 |
"88": "SUCCESS",
|
| 248 |
"89": "NON_PARTICIPATING",
|
| 249 |
"9": "NOT_ALIVE",
|
|
|
|
| 253 |
"93": "SUCCESS",
|
| 254 |
"94": "SUCCESS",
|
| 255 |
"95": "NON_PARTICIPATING",
|
| 256 |
+
"96": "NON_PARTICIPATING",
|
| 257 |
"97": "NON_PARTICIPATING",
|
| 258 |
"98": "SUCCESS",
|
| 259 |
"99": "SUCCESS"
|
|
|
|
| 264 |
"attention_bias": false,
|
| 265 |
"attention_dropout": 0.0,
|
| 266 |
"block_list": [
|
| 267 |
+
6447400,
|
| 268 |
+
6447420,
|
| 269 |
+
6447429
|
| 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": 4,
|
| 277 |
"intermediate_size": 5632,
|
| 278 |
+
"last_allreduce_block": 6447169,
|
| 279 |
"max_position_embeddings": 2048,
|
| 280 |
"mlp_bias": false,
|
| 281 |
"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 sha256:91799e4e92d7d38acd20ff46c5f84c0aceffb2a385815fd5f9cccbec86af5f31
|
| 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:bc143fb7bb03e2eb633e4f6795f5766094051914dfd394c450d53892a77a5184
|
| 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 sha256:646ad62ced6d38eb20e0316c8badfc843bf222485033d284ae1eb805c203d4f2
|
| 3 |
size 4400216536
|