Run 5. Outer Step 112. Inner Step 2.
Browse files- config.json +73 -73
- inner_optimizer.pt +1 -1
- model.safetensors +1 -1
config.json
CHANGED
|
@@ -5,9 +5,9 @@
|
|
| 5 |
"1": "NON_PARTICIPATING",
|
| 6 |
"10": "NOT_ALIVE",
|
| 7 |
"100": "NOT_ALIVE",
|
| 8 |
-
"101": "
|
| 9 |
"102": "NON_PARTICIPATING",
|
| 10 |
-
"103": "
|
| 11 |
"104": "NON_PARTICIPATING",
|
| 12 |
"105": "NON_PARTICIPATING",
|
| 13 |
"106": "NON_PARTICIPATING",
|
|
@@ -15,8 +15,8 @@
|
|
| 15 |
"108": "NOT_ALIVE",
|
| 16 |
"109": "NOT_ALIVE",
|
| 17 |
"11": "NOT_ALIVE",
|
| 18 |
-
"110": "
|
| 19 |
-
"111": "
|
| 20 |
"112": "NON_PARTICIPATING",
|
| 21 |
"113": "NOT_ALIVE",
|
| 22 |
"114": "NON_PARTICIPATING",
|
|
@@ -39,52 +39,52 @@
|
|
| 39 |
"13": "NOT_ALIVE",
|
| 40 |
"130": "NON_PARTICIPATING",
|
| 41 |
"131": "NOT_ALIVE",
|
| 42 |
-
"132": "
|
| 43 |
"133": "SUCCESS",
|
| 44 |
"134": "NON_PARTICIPATING",
|
| 45 |
"135": "NON_PARTICIPATING",
|
| 46 |
-
"136": "
|
| 47 |
"137": "NON_PARTICIPATING",
|
| 48 |
-
"138": "
|
| 49 |
"139": "NON_PARTICIPATING",
|
| 50 |
-
"14": "
|
| 51 |
"140": "NOT_ALIVE",
|
| 52 |
"141": "NOT_ALIVE",
|
| 53 |
"142": "NON_PARTICIPATING",
|
| 54 |
"143": "NON_PARTICIPATING",
|
| 55 |
"144": "NON_PARTICIPATING",
|
| 56 |
"145": "NON_PARTICIPATING",
|
| 57 |
-
"146": "
|
| 58 |
-
"147": "
|
| 59 |
"148": "NON_PARTICIPATING",
|
| 60 |
"149": "NON_PARTICIPATING",
|
| 61 |
"15": "NOT_ALIVE",
|
| 62 |
"150": "NON_PARTICIPATING",
|
| 63 |
"151": "NOT_ALIVE",
|
| 64 |
"152": "NOT_ALIVE",
|
| 65 |
-
"153": "
|
| 66 |
-
"154": "
|
| 67 |
-
"155": "
|
| 68 |
"156": "NOT_ALIVE",
|
| 69 |
"157": "NOT_ALIVE",
|
| 70 |
-
"158": "
|
| 71 |
-
"159": "
|
| 72 |
"16": "NON_PARTICIPATING",
|
| 73 |
"160": "NON_PARTICIPATING",
|
| 74 |
"161": "NON_PARTICIPATING",
|
| 75 |
-
"162": "
|
| 76 |
"163": "NOT_ALIVE",
|
| 77 |
-
"164": "
|
| 78 |
"165": "FAIL",
|
| 79 |
-
"166": "
|
| 80 |
"167": "NOT_ALIVE",
|
| 81 |
-
"168": "
|
| 82 |
"169": "NOT_ALIVE",
|
| 83 |
"17": "NOT_ALIVE",
|
| 84 |
"170": "NON_PARTICIPATING",
|
| 85 |
"171": "NOT_ALIVE",
|
| 86 |
"172": "NON_PARTICIPATING",
|
| 87 |
-
"173": "
|
| 88 |
"174": "NOT_ALIVE",
|
| 89 |
"175": "NOT_ALIVE",
|
| 90 |
"176": "NOT_ALIVE",
|
|
@@ -95,20 +95,20 @@
|
|
| 95 |
"180": "NOT_ALIVE",
|
| 96 |
"181": "NOT_ALIVE",
|
| 97 |
"182": "NOT_ALIVE",
|
| 98 |
-
"183": "
|
| 99 |
-
"184": "
|
| 100 |
"185": "NON_PARTICIPATING",
|
| 101 |
"186": "NON_PARTICIPATING",
|
| 102 |
-
"187": "
|
| 103 |
"188": "NON_PARTICIPATING",
|
| 104 |
"189": "NOT_ALIVE",
|
| 105 |
"19": "NOT_ALIVE",
|
| 106 |
"190": "NON_PARTICIPATING",
|
| 107 |
"191": "NOT_ALIVE",
|
| 108 |
"192": "NOT_ALIVE",
|
| 109 |
-
"193": "
|
| 110 |
-
"194": "
|
| 111 |
-
"195": "
|
| 112 |
"196": "NOT_ALIVE",
|
| 113 |
"197": "NOT_ALIVE",
|
| 114 |
"198": "NON_PARTICIPATING",
|
|
@@ -123,99 +123,99 @@
|
|
| 123 |
"205": "NON_PARTICIPATING",
|
| 124 |
"206": "NON_PARTICIPATING",
|
| 125 |
"207": "NOT_ALIVE",
|
| 126 |
-
"208": "
|
| 127 |
"209": "NOT_ALIVE",
|
| 128 |
"21": "NOT_ALIVE",
|
| 129 |
"210": "NON_PARTICIPATING",
|
| 130 |
"211": "NON_PARTICIPATING",
|
| 131 |
-
"212": "
|
| 132 |
-
"213": "
|
| 133 |
-
"214": "
|
| 134 |
-
"215": "
|
| 135 |
-
"216": "
|
| 136 |
"217": "NON_PARTICIPATING",
|
| 137 |
"218": "NON_PARTICIPATING",
|
| 138 |
"219": "NON_PARTICIPATING",
|
| 139 |
-
"22": "
|
| 140 |
"220": "NON_PARTICIPATING",
|
| 141 |
"221": "NOT_ALIVE",
|
| 142 |
"222": "NOT_ALIVE",
|
| 143 |
-
"223": "
|
| 144 |
"224": "NOT_ALIVE",
|
| 145 |
"225": "NOT_ALIVE",
|
| 146 |
"226": "NOT_ALIVE",
|
| 147 |
"227": "NON_PARTICIPATING",
|
| 148 |
"228": "NON_PARTICIPATING",
|
| 149 |
"229": "NON_PARTICIPATING",
|
| 150 |
-
"23": "
|
| 151 |
"230": "NOT_ALIVE",
|
| 152 |
"231": "NON_PARTICIPATING",
|
| 153 |
"232": "NOT_ALIVE",
|
| 154 |
-
"233": "
|
| 155 |
"234": "NON_PARTICIPATING",
|
| 156 |
"235": "NOT_ALIVE",
|
| 157 |
"236": "NOT_ALIVE",
|
| 158 |
"237": "SUCCESS",
|
| 159 |
-
"238": "
|
| 160 |
"239": "NOT_ALIVE",
|
| 161 |
-
"24": "
|
| 162 |
"240": "NOT_ALIVE",
|
| 163 |
"241": "NOT_ALIVE",
|
| 164 |
"242": "NON_PARTICIPATING",
|
| 165 |
"243": "NOT_ALIVE",
|
| 166 |
"244": "NOT_ALIVE",
|
| 167 |
-
"245": "
|
| 168 |
"246": "NON_PARTICIPATING",
|
| 169 |
-
"247": "
|
| 170 |
"248": "NOT_ALIVE",
|
| 171 |
"249": "NOT_ALIVE",
|
| 172 |
"25": "SUCCESS",
|
| 173 |
-
"250": "
|
| 174 |
"251": "NOT_ALIVE",
|
| 175 |
-
"252": "
|
| 176 |
"253": "NOT_ALIVE",
|
| 177 |
-
"254": "
|
| 178 |
"255": "NON_PARTICIPATING",
|
| 179 |
-
"26": "
|
| 180 |
-
"27": "
|
| 181 |
"28": "NOT_ALIVE",
|
| 182 |
"29": "NON_PARTICIPATING",
|
| 183 |
"3": "NON_PARTICIPATING",
|
| 184 |
-
"30": "
|
| 185 |
"31": "NON_PARTICIPATING",
|
| 186 |
"32": "FAIL",
|
| 187 |
"33": "NOT_ALIVE",
|
| 188 |
"34": "NOT_ALIVE",
|
| 189 |
-
"35": "
|
| 190 |
"36": "NOT_ALIVE",
|
| 191 |
"37": "NOT_ALIVE",
|
| 192 |
"38": "NON_PARTICIPATING",
|
| 193 |
"39": "FAIL",
|
| 194 |
-
"4": "
|
| 195 |
-
"40": "
|
| 196 |
-
"41": "
|
| 197 |
-
"42": "
|
| 198 |
-
"43": "
|
| 199 |
"44": "NON_PARTICIPATING",
|
| 200 |
"45": "NOT_ALIVE",
|
| 201 |
"46": "NOT_ALIVE",
|
| 202 |
"47": "NON_PARTICIPATING",
|
| 203 |
"48": "NON_PARTICIPATING",
|
| 204 |
-
"49": "
|
| 205 |
-
"5": "
|
| 206 |
-
"50": "
|
| 207 |
"51": "NON_PARTICIPATING",
|
| 208 |
"52": "NON_PARTICIPATING",
|
| 209 |
"53": "NON_PARTICIPATING",
|
| 210 |
"54": "NON_PARTICIPATING",
|
| 211 |
"55": "NOT_ALIVE",
|
| 212 |
-
"56": "
|
| 213 |
"57": "NON_PARTICIPATING",
|
| 214 |
-
"58": "
|
| 215 |
-
"59": "
|
| 216 |
-
"6": "
|
| 217 |
-
"60": "
|
| 218 |
-
"61": "
|
| 219 |
"62": "NON_PARTICIPATING",
|
| 220 |
"63": "NON_PARTICIPATING",
|
| 221 |
"64": "NOT_ALIVE",
|
|
@@ -224,11 +224,11 @@
|
|
| 224 |
"67": "NOT_ALIVE",
|
| 225 |
"68": "NON_PARTICIPATING",
|
| 226 |
"69": "NOT_ALIVE",
|
| 227 |
-
"7": "
|
| 228 |
"70": "NON_PARTICIPATING",
|
| 229 |
-
"71": "
|
| 230 |
"72": "NON_PARTICIPATING",
|
| 231 |
-
"73": "
|
| 232 |
"74": "NON_PARTICIPATING",
|
| 233 |
"75": "NON_PARTICIPATING",
|
| 234 |
"76": "NON_PARTICIPATING",
|
|
@@ -238,7 +238,7 @@
|
|
| 238 |
"8": "NOT_ALIVE",
|
| 239 |
"80": "NON_PARTICIPATING",
|
| 240 |
"81": "NON_PARTICIPATING",
|
| 241 |
-
"82": "
|
| 242 |
"83": "NON_PARTICIPATING",
|
| 243 |
"84": "NON_PARTICIPATING",
|
| 244 |
"85": "NON_PARTICIPATING",
|
|
@@ -249,12 +249,12 @@
|
|
| 249 |
"9": "NOT_ALIVE",
|
| 250 |
"90": "NON_PARTICIPATING",
|
| 251 |
"91": "NOT_ALIVE",
|
| 252 |
-
"92": "
|
| 253 |
"93": "NON_PARTICIPATING",
|
| 254 |
-
"94": "
|
| 255 |
"95": "NOT_ALIVE",
|
| 256 |
-
"96": "
|
| 257 |
-
"97": "
|
| 258 |
"98": "NON_PARTICIPATING",
|
| 259 |
"99": "NOT_ALIVE"
|
| 260 |
},
|
|
@@ -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",
|
|
|
|
| 5 |
"1": "NON_PARTICIPATING",
|
| 6 |
"10": "NOT_ALIVE",
|
| 7 |
"100": "NOT_ALIVE",
|
| 8 |
+
"101": "NOT_ALIVE",
|
| 9 |
"102": "NON_PARTICIPATING",
|
| 10 |
+
"103": "NON_PARTICIPATING",
|
| 11 |
"104": "NON_PARTICIPATING",
|
| 12 |
"105": "NON_PARTICIPATING",
|
| 13 |
"106": "NON_PARTICIPATING",
|
|
|
|
| 15 |
"108": "NOT_ALIVE",
|
| 16 |
"109": "NOT_ALIVE",
|
| 17 |
"11": "NOT_ALIVE",
|
| 18 |
+
"110": "NOT_ALIVE",
|
| 19 |
+
"111": "NOT_ALIVE",
|
| 20 |
"112": "NON_PARTICIPATING",
|
| 21 |
"113": "NOT_ALIVE",
|
| 22 |
"114": "NON_PARTICIPATING",
|
|
|
|
| 39 |
"13": "NOT_ALIVE",
|
| 40 |
"130": "NON_PARTICIPATING",
|
| 41 |
"131": "NOT_ALIVE",
|
| 42 |
+
"132": "NON_PARTICIPATING",
|
| 43 |
"133": "SUCCESS",
|
| 44 |
"134": "NON_PARTICIPATING",
|
| 45 |
"135": "NON_PARTICIPATING",
|
| 46 |
+
"136": "NON_PARTICIPATING",
|
| 47 |
"137": "NON_PARTICIPATING",
|
| 48 |
+
"138": "NON_PARTICIPATING",
|
| 49 |
"139": "NON_PARTICIPATING",
|
| 50 |
+
"14": "NON_PARTICIPATING",
|
| 51 |
"140": "NOT_ALIVE",
|
| 52 |
"141": "NOT_ALIVE",
|
| 53 |
"142": "NON_PARTICIPATING",
|
| 54 |
"143": "NON_PARTICIPATING",
|
| 55 |
"144": "NON_PARTICIPATING",
|
| 56 |
"145": "NON_PARTICIPATING",
|
| 57 |
+
"146": "NOT_ALIVE",
|
| 58 |
+
"147": "NON_PARTICIPATING",
|
| 59 |
"148": "NON_PARTICIPATING",
|
| 60 |
"149": "NON_PARTICIPATING",
|
| 61 |
"15": "NOT_ALIVE",
|
| 62 |
"150": "NON_PARTICIPATING",
|
| 63 |
"151": "NOT_ALIVE",
|
| 64 |
"152": "NOT_ALIVE",
|
| 65 |
+
"153": "NON_PARTICIPATING",
|
| 66 |
+
"154": "FAIL",
|
| 67 |
+
"155": "NOT_ALIVE",
|
| 68 |
"156": "NOT_ALIVE",
|
| 69 |
"157": "NOT_ALIVE",
|
| 70 |
+
"158": "NOT_ALIVE",
|
| 71 |
+
"159": "FAIL",
|
| 72 |
"16": "NON_PARTICIPATING",
|
| 73 |
"160": "NON_PARTICIPATING",
|
| 74 |
"161": "NON_PARTICIPATING",
|
| 75 |
+
"162": "NOT_ALIVE",
|
| 76 |
"163": "NOT_ALIVE",
|
| 77 |
+
"164": "SUCCESS",
|
| 78 |
"165": "FAIL",
|
| 79 |
+
"166": "NON_PARTICIPATING",
|
| 80 |
"167": "NOT_ALIVE",
|
| 81 |
+
"168": "NON_PARTICIPATING",
|
| 82 |
"169": "NOT_ALIVE",
|
| 83 |
"17": "NOT_ALIVE",
|
| 84 |
"170": "NON_PARTICIPATING",
|
| 85 |
"171": "NOT_ALIVE",
|
| 86 |
"172": "NON_PARTICIPATING",
|
| 87 |
+
"173": "SUCCESS",
|
| 88 |
"174": "NOT_ALIVE",
|
| 89 |
"175": "NOT_ALIVE",
|
| 90 |
"176": "NOT_ALIVE",
|
|
|
|
| 95 |
"180": "NOT_ALIVE",
|
| 96 |
"181": "NOT_ALIVE",
|
| 97 |
"182": "NOT_ALIVE",
|
| 98 |
+
"183": "NON_PARTICIPATING",
|
| 99 |
+
"184": "NOT_ALIVE",
|
| 100 |
"185": "NON_PARTICIPATING",
|
| 101 |
"186": "NON_PARTICIPATING",
|
| 102 |
+
"187": "NON_PARTICIPATING",
|
| 103 |
"188": "NON_PARTICIPATING",
|
| 104 |
"189": "NOT_ALIVE",
|
| 105 |
"19": "NOT_ALIVE",
|
| 106 |
"190": "NON_PARTICIPATING",
|
| 107 |
"191": "NOT_ALIVE",
|
| 108 |
"192": "NOT_ALIVE",
|
| 109 |
+
"193": "NOT_ALIVE",
|
| 110 |
+
"194": "NON_PARTICIPATING",
|
| 111 |
+
"195": "NON_PARTICIPATING",
|
| 112 |
"196": "NOT_ALIVE",
|
| 113 |
"197": "NOT_ALIVE",
|
| 114 |
"198": "NON_PARTICIPATING",
|
|
|
|
| 123 |
"205": "NON_PARTICIPATING",
|
| 124 |
"206": "NON_PARTICIPATING",
|
| 125 |
"207": "NOT_ALIVE",
|
| 126 |
+
"208": "NON_PARTICIPATING",
|
| 127 |
"209": "NOT_ALIVE",
|
| 128 |
"21": "NOT_ALIVE",
|
| 129 |
"210": "NON_PARTICIPATING",
|
| 130 |
"211": "NON_PARTICIPATING",
|
| 131 |
+
"212": "SUCCESS",
|
| 132 |
+
"213": "NON_PARTICIPATING",
|
| 133 |
+
"214": "NON_PARTICIPATING",
|
| 134 |
+
"215": "FAIL",
|
| 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": "NOT_ALIVE",
|
| 142 |
"222": "NOT_ALIVE",
|
| 143 |
+
"223": "NON_PARTICIPATING",
|
| 144 |
"224": "NOT_ALIVE",
|
| 145 |
"225": "NOT_ALIVE",
|
| 146 |
"226": "NOT_ALIVE",
|
| 147 |
"227": "NON_PARTICIPATING",
|
| 148 |
"228": "NON_PARTICIPATING",
|
| 149 |
"229": "NON_PARTICIPATING",
|
| 150 |
+
"23": "NOT_ALIVE",
|
| 151 |
"230": "NOT_ALIVE",
|
| 152 |
"231": "NON_PARTICIPATING",
|
| 153 |
"232": "NOT_ALIVE",
|
| 154 |
+
"233": "NOT_ALIVE",
|
| 155 |
"234": "NON_PARTICIPATING",
|
| 156 |
"235": "NOT_ALIVE",
|
| 157 |
"236": "NOT_ALIVE",
|
| 158 |
"237": "SUCCESS",
|
| 159 |
+
"238": "NOT_ALIVE",
|
| 160 |
"239": "NOT_ALIVE",
|
| 161 |
+
"24": "NON_PARTICIPATING",
|
| 162 |
"240": "NOT_ALIVE",
|
| 163 |
"241": "NOT_ALIVE",
|
| 164 |
"242": "NON_PARTICIPATING",
|
| 165 |
"243": "NOT_ALIVE",
|
| 166 |
"244": "NOT_ALIVE",
|
| 167 |
+
"245": "NOT_ALIVE",
|
| 168 |
"246": "NON_PARTICIPATING",
|
| 169 |
+
"247": "NON_PARTICIPATING",
|
| 170 |
"248": "NOT_ALIVE",
|
| 171 |
"249": "NOT_ALIVE",
|
| 172 |
"25": "SUCCESS",
|
| 173 |
+
"250": "SUCCESS",
|
| 174 |
"251": "NOT_ALIVE",
|
| 175 |
+
"252": "NON_PARTICIPATING",
|
| 176 |
"253": "NOT_ALIVE",
|
| 177 |
+
"254": "NOT_ALIVE",
|
| 178 |
"255": "NON_PARTICIPATING",
|
| 179 |
+
"26": "NON_PARTICIPATING",
|
| 180 |
+
"27": "NON_PARTICIPATING",
|
| 181 |
"28": "NOT_ALIVE",
|
| 182 |
"29": "NON_PARTICIPATING",
|
| 183 |
"3": "NON_PARTICIPATING",
|
| 184 |
+
"30": "NON_PARTICIPATING",
|
| 185 |
"31": "NON_PARTICIPATING",
|
| 186 |
"32": "FAIL",
|
| 187 |
"33": "NOT_ALIVE",
|
| 188 |
"34": "NOT_ALIVE",
|
| 189 |
+
"35": "SUCCESS",
|
| 190 |
"36": "NOT_ALIVE",
|
| 191 |
"37": "NOT_ALIVE",
|
| 192 |
"38": "NON_PARTICIPATING",
|
| 193 |
"39": "FAIL",
|
| 194 |
+
"4": "NOT_ALIVE",
|
| 195 |
+
"40": "NOT_ALIVE",
|
| 196 |
+
"41": "SUCCESS",
|
| 197 |
+
"42": "NOT_ALIVE",
|
| 198 |
+
"43": "NON_PARTICIPATING",
|
| 199 |
"44": "NON_PARTICIPATING",
|
| 200 |
"45": "NOT_ALIVE",
|
| 201 |
"46": "NOT_ALIVE",
|
| 202 |
"47": "NON_PARTICIPATING",
|
| 203 |
"48": "NON_PARTICIPATING",
|
| 204 |
+
"49": "NON_PARTICIPATING",
|
| 205 |
+
"5": "NON_PARTICIPATING",
|
| 206 |
+
"50": "NOT_ALIVE",
|
| 207 |
"51": "NON_PARTICIPATING",
|
| 208 |
"52": "NON_PARTICIPATING",
|
| 209 |
"53": "NON_PARTICIPATING",
|
| 210 |
"54": "NON_PARTICIPATING",
|
| 211 |
"55": "NOT_ALIVE",
|
| 212 |
+
"56": "NON_PARTICIPATING",
|
| 213 |
"57": "NON_PARTICIPATING",
|
| 214 |
+
"58": "SUCCESS",
|
| 215 |
+
"59": "NON_PARTICIPATING",
|
| 216 |
+
"6": "NOT_ALIVE",
|
| 217 |
+
"60": "NOT_ALIVE",
|
| 218 |
+
"61": "NOT_ALIVE",
|
| 219 |
"62": "NON_PARTICIPATING",
|
| 220 |
"63": "NON_PARTICIPATING",
|
| 221 |
"64": "NOT_ALIVE",
|
|
|
|
| 224 |
"67": "NOT_ALIVE",
|
| 225 |
"68": "NON_PARTICIPATING",
|
| 226 |
"69": "NOT_ALIVE",
|
| 227 |
+
"7": "NOT_ALIVE",
|
| 228 |
"70": "NON_PARTICIPATING",
|
| 229 |
+
"71": "NON_PARTICIPATING",
|
| 230 |
"72": "NON_PARTICIPATING",
|
| 231 |
+
"73": "NOT_ALIVE",
|
| 232 |
"74": "NON_PARTICIPATING",
|
| 233 |
"75": "NON_PARTICIPATING",
|
| 234 |
"76": "NON_PARTICIPATING",
|
|
|
|
| 238 |
"8": "NOT_ALIVE",
|
| 239 |
"80": "NON_PARTICIPATING",
|
| 240 |
"81": "NON_PARTICIPATING",
|
| 241 |
+
"82": "NON_PARTICIPATING",
|
| 242 |
"83": "NON_PARTICIPATING",
|
| 243 |
"84": "NON_PARTICIPATING",
|
| 244 |
"85": "NON_PARTICIPATING",
|
|
|
|
| 249 |
"9": "NOT_ALIVE",
|
| 250 |
"90": "NON_PARTICIPATING",
|
| 251 |
"91": "NOT_ALIVE",
|
| 252 |
+
"92": "FAIL",
|
| 253 |
"93": "NON_PARTICIPATING",
|
| 254 |
+
"94": "FAIL",
|
| 255 |
"95": "NOT_ALIVE",
|
| 256 |
+
"96": "NON_PARTICIPATING",
|
| 257 |
+
"97": "NOT_ALIVE",
|
| 258 |
"98": "NON_PARTICIPATING",
|
| 259 |
"99": "NOT_ALIVE"
|
| 260 |
},
|
|
|
|
| 264 |
"attention_bias": false,
|
| 265 |
"attention_dropout": 0.0,
|
| 266 |
"block_list": [
|
| 267 |
+
5971200,
|
| 268 |
+
5971265
|
| 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": 2,
|
| 276 |
"intermediate_size": 5632,
|
| 277 |
+
"last_allreduce_block": 5971084,
|
| 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 8800564076
|
|
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:5355b084e82ffa18bf02f871853646a3d1b507e3ebaf3dc57c0777040b346d28
|
| 3 |
size 8800564076
|
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 sha256:658bf0befa665cd56d2d611f9cb94e4e98cc45ca944cce890adb8928ea5b855d
|
| 3 |
size 4400216536
|