Run 5. Outer Step 59. Inner Step 3.
Browse files- config.json +68 -68
- inner_optimizer.pt +1 -1
- model.safetensors +1 -1
config.json
CHANGED
|
@@ -7,7 +7,7 @@
|
|
| 7 |
"100": "NOT_ALIVE",
|
| 8 |
"101": "NON_PARTICIPATING",
|
| 9 |
"102": "NOT_ALIVE",
|
| 10 |
-
"103": "
|
| 11 |
"104": "NOT_ALIVE",
|
| 12 |
"105": "SUCCESS",
|
| 13 |
"106": "NOT_ALIVE",
|
|
@@ -18,46 +18,46 @@
|
|
| 18 |
"110": "SUCCESS",
|
| 19 |
"111": "SUCCESS",
|
| 20 |
"112": "NOT_ALIVE",
|
| 21 |
-
"113": "
|
| 22 |
-
"114": "
|
| 23 |
"115": "NOT_ALIVE",
|
| 24 |
"116": "NON_PARTICIPATING",
|
| 25 |
"117": "NOT_ALIVE",
|
| 26 |
-
"118": "
|
| 27 |
-
"119": "
|
| 28 |
"12": "NOT_ALIVE",
|
| 29 |
-
"120": "
|
| 30 |
-
"121": "
|
| 31 |
"122": "SUCCESS",
|
| 32 |
-
"123": "
|
| 33 |
"124": "SUCCESS",
|
| 34 |
"125": "NOT_ALIVE",
|
| 35 |
"126": "NOT_ALIVE",
|
| 36 |
"127": "NON_PARTICIPATING",
|
| 37 |
"128": "NOT_ALIVE",
|
| 38 |
-
"129": "
|
| 39 |
-
"13": "
|
| 40 |
"130": "SUCCESS",
|
| 41 |
"131": "NOT_ALIVE",
|
| 42 |
"132": "NOT_ALIVE",
|
| 43 |
"133": "SUCCESS",
|
| 44 |
-
"134": "
|
| 45 |
"135": "NON_PARTICIPATING",
|
| 46 |
"136": "SUCCESS",
|
| 47 |
"137": "NOT_ALIVE",
|
| 48 |
"138": "NOT_ALIVE",
|
| 49 |
-
"139": "
|
| 50 |
"14": "NON_PARTICIPATING",
|
| 51 |
"140": "NOT_ALIVE",
|
| 52 |
-
"141": "
|
| 53 |
"142": "SUCCESS",
|
| 54 |
"143": "NON_PARTICIPATING",
|
| 55 |
"144": "NON_PARTICIPATING",
|
| 56 |
-
"145": "
|
| 57 |
"146": "NOT_ALIVE",
|
| 58 |
-
"147": "
|
| 59 |
"148": "SUCCESS",
|
| 60 |
-
"149": "
|
| 61 |
"15": "NON_PARTICIPATING",
|
| 62 |
"150": "SUCCESS",
|
| 63 |
"151": "NOT_ALIVE",
|
|
@@ -71,16 +71,16 @@
|
|
| 71 |
"159": "NOT_ALIVE",
|
| 72 |
"16": "NON_PARTICIPATING",
|
| 73 |
"160": "NON_PARTICIPATING",
|
| 74 |
-
"161": "
|
| 75 |
"162": "NON_PARTICIPATING",
|
| 76 |
"163": "NOT_ALIVE",
|
| 77 |
"164": "SUCCESS",
|
| 78 |
"165": "SUCCESS",
|
| 79 |
-
"166": "
|
| 80 |
"167": "NON_PARTICIPATING",
|
| 81 |
-
"168": "
|
| 82 |
-
"169": "
|
| 83 |
-
"17": "
|
| 84 |
"170": "NON_PARTICIPATING",
|
| 85 |
"171": "SUCCESS",
|
| 86 |
"172": "NOT_ALIVE",
|
|
@@ -106,52 +106,52 @@
|
|
| 106 |
"190": "NON_PARTICIPATING",
|
| 107 |
"191": "NOT_ALIVE",
|
| 108 |
"192": "SUCCESS",
|
| 109 |
-
"193": "
|
| 110 |
"194": "NOT_ALIVE",
|
| 111 |
"195": "NOT_ALIVE",
|
| 112 |
"196": "NOT_ALIVE",
|
| 113 |
-
"197": "
|
| 114 |
"198": "NOT_ALIVE",
|
| 115 |
-
"199": "
|
| 116 |
"2": "NOT_ALIVE",
|
| 117 |
"20": "NON_PARTICIPATING",
|
| 118 |
"200": "NON_PARTICIPATING",
|
| 119 |
"201": "NON_PARTICIPATING",
|
| 120 |
"202": "NON_PARTICIPATING",
|
| 121 |
"203": "NON_PARTICIPATING",
|
| 122 |
-
"204": "
|
| 123 |
"205": "NOT_ALIVE",
|
| 124 |
"206": "SUCCESS",
|
| 125 |
"207": "NOT_ALIVE",
|
| 126 |
-
"208": "
|
| 127 |
"209": "SUCCESS",
|
| 128 |
"21": "SUCCESS",
|
| 129 |
"210": "SUCCESS",
|
| 130 |
"211": "NOT_ALIVE",
|
| 131 |
"212": "NOT_ALIVE",
|
| 132 |
-
"213": "
|
| 133 |
"214": "NON_PARTICIPATING",
|
| 134 |
-
"215": "
|
| 135 |
"216": "NOT_ALIVE",
|
| 136 |
"217": "NON_PARTICIPATING",
|
| 137 |
-
"218": "
|
| 138 |
"219": "NOT_ALIVE",
|
| 139 |
"22": "NON_PARTICIPATING",
|
| 140 |
-
"220": "
|
| 141 |
"221": "SUCCESS",
|
| 142 |
"222": "NON_PARTICIPATING",
|
| 143 |
-
"223": "
|
| 144 |
"224": "NOT_ALIVE",
|
| 145 |
-
"225": "
|
| 146 |
"226": "NOT_ALIVE",
|
| 147 |
-
"227": "
|
| 148 |
"228": "NOT_ALIVE",
|
| 149 |
"229": "NOT_ALIVE",
|
| 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": "NON_PARTICIPATING",
|
|
@@ -159,21 +159,21 @@
|
|
| 159 |
"238": "NON_PARTICIPATING",
|
| 160 |
"239": "NON_PARTICIPATING",
|
| 161 |
"24": "SUCCESS",
|
| 162 |
-
"240": "
|
| 163 |
"241": "NOT_ALIVE",
|
| 164 |
"242": "SUCCESS",
|
| 165 |
"243": "NON_PARTICIPATING",
|
| 166 |
"244": "NOT_ALIVE",
|
| 167 |
"245": "NOT_ALIVE",
|
| 168 |
"246": "NON_PARTICIPATING",
|
| 169 |
-
"247": "
|
| 170 |
"248": "NON_PARTICIPATING",
|
| 171 |
-
"249": "
|
| 172 |
"25": "SUCCESS",
|
| 173 |
"250": "NON_PARTICIPATING",
|
| 174 |
"251": "NOT_ALIVE",
|
| 175 |
-
"252": "
|
| 176 |
-
"253": "
|
| 177 |
"254": "NOT_ALIVE",
|
| 178 |
"255": "NON_PARTICIPATING",
|
| 179 |
"26": "SUCCESS",
|
|
@@ -181,11 +181,11 @@
|
|
| 181 |
"28": "NOT_ALIVE",
|
| 182 |
"29": "NOT_ALIVE",
|
| 183 |
"3": "SUCCESS",
|
| 184 |
-
"30": "
|
| 185 |
"31": "NON_PARTICIPATING",
|
| 186 |
-
"32": "
|
| 187 |
"33": "NOT_ALIVE",
|
| 188 |
-
"34": "
|
| 189 |
"35": "NOT_ALIVE",
|
| 190 |
"36": "NON_PARTICIPATING",
|
| 191 |
"37": "NON_PARTICIPATING",
|
|
@@ -193,49 +193,49 @@
|
|
| 193 |
"39": "SUCCESS",
|
| 194 |
"4": "NOT_ALIVE",
|
| 195 |
"40": "SUCCESS",
|
| 196 |
-
"41": "
|
| 197 |
"42": "SUCCESS",
|
| 198 |
-
"43": "
|
| 199 |
"44": "NOT_ALIVE",
|
| 200 |
-
"45": "
|
| 201 |
-
"46": "
|
| 202 |
-
"47": "
|
| 203 |
"48": "SUCCESS",
|
| 204 |
"49": "NOT_ALIVE",
|
| 205 |
-
"5": "
|
| 206 |
"50": "NON_PARTICIPATING",
|
| 207 |
"51": "NOT_ALIVE",
|
| 208 |
"52": "NOT_ALIVE",
|
| 209 |
-
"53": "
|
| 210 |
"54": "SUCCESS",
|
| 211 |
-
"55": "
|
| 212 |
"56": "SUCCESS",
|
| 213 |
"57": "SUCCESS",
|
| 214 |
"58": "SUCCESS",
|
| 215 |
"59": "NOT_ALIVE",
|
| 216 |
-
"6": "
|
| 217 |
"60": "NON_PARTICIPATING",
|
| 218 |
"61": "NOT_ALIVE",
|
| 219 |
"62": "NOT_ALIVE",
|
| 220 |
"63": "NON_PARTICIPATING",
|
| 221 |
"64": "NON_PARTICIPATING",
|
| 222 |
-
"65": "
|
| 223 |
-
"66": "
|
| 224 |
-
"67": "
|
| 225 |
-
"68": "
|
| 226 |
"69": "SUCCESS",
|
| 227 |
"7": "SUCCESS",
|
| 228 |
"70": "NOT_ALIVE",
|
| 229 |
"71": "NON_PARTICIPATING",
|
| 230 |
-
"72": "
|
| 231 |
"73": "SUCCESS",
|
| 232 |
"74": "SUCCESS",
|
| 233 |
"75": "SUCCESS",
|
| 234 |
"76": "SUCCESS",
|
| 235 |
"77": "NOT_ALIVE",
|
| 236 |
-
"78": "
|
| 237 |
"79": "NON_PARTICIPATING",
|
| 238 |
-
"8": "
|
| 239 |
"80": "NON_PARTICIPATING",
|
| 240 |
"81": "NOT_ALIVE",
|
| 241 |
"82": "NOT_ALIVE",
|
|
@@ -244,19 +244,19 @@
|
|
| 244 |
"85": "NON_PARTICIPATING",
|
| 245 |
"86": "SUCCESS",
|
| 246 |
"87": "NON_PARTICIPATING",
|
| 247 |
-
"88": "
|
| 248 |
-
"89": "
|
| 249 |
"9": "NOT_ALIVE",
|
| 250 |
"90": "SUCCESS",
|
| 251 |
"91": "NOT_ALIVE",
|
| 252 |
"92": "SUCCESS",
|
| 253 |
"93": "NON_PARTICIPATING",
|
| 254 |
-
"94": "
|
| 255 |
-
"95": "
|
| 256 |
"96": "NON_PARTICIPATING",
|
| 257 |
"97": "SUCCESS",
|
| 258 |
"98": "SUCCESS",
|
| 259 |
-
"99": "
|
| 260 |
},
|
| 261 |
"architectures": [
|
| 262 |
"LlamaForCausalLM"
|
|
@@ -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",
|
|
|
|
| 7 |
"100": "NOT_ALIVE",
|
| 8 |
"101": "NON_PARTICIPATING",
|
| 9 |
"102": "NOT_ALIVE",
|
| 10 |
+
"103": "SUCCESS",
|
| 11 |
"104": "NOT_ALIVE",
|
| 12 |
"105": "SUCCESS",
|
| 13 |
"106": "NOT_ALIVE",
|
|
|
|
| 18 |
"110": "SUCCESS",
|
| 19 |
"111": "SUCCESS",
|
| 20 |
"112": "NOT_ALIVE",
|
| 21 |
+
"113": "NOT_ALIVE",
|
| 22 |
+
"114": "NOT_ALIVE",
|
| 23 |
"115": "NOT_ALIVE",
|
| 24 |
"116": "NON_PARTICIPATING",
|
| 25 |
"117": "NOT_ALIVE",
|
| 26 |
+
"118": "SUCCESS",
|
| 27 |
+
"119": "NON_PARTICIPATING",
|
| 28 |
"12": "NOT_ALIVE",
|
| 29 |
+
"120": "NON_PARTICIPATING",
|
| 30 |
+
"121": "NON_PARTICIPATING",
|
| 31 |
"122": "SUCCESS",
|
| 32 |
+
"123": "NOT_ALIVE",
|
| 33 |
"124": "SUCCESS",
|
| 34 |
"125": "NOT_ALIVE",
|
| 35 |
"126": "NOT_ALIVE",
|
| 36 |
"127": "NON_PARTICIPATING",
|
| 37 |
"128": "NOT_ALIVE",
|
| 38 |
+
"129": "NON_PARTICIPATING",
|
| 39 |
+
"13": "NON_PARTICIPATING",
|
| 40 |
"130": "SUCCESS",
|
| 41 |
"131": "NOT_ALIVE",
|
| 42 |
"132": "NOT_ALIVE",
|
| 43 |
"133": "SUCCESS",
|
| 44 |
+
"134": "NOT_ALIVE",
|
| 45 |
"135": "NON_PARTICIPATING",
|
| 46 |
"136": "SUCCESS",
|
| 47 |
"137": "NOT_ALIVE",
|
| 48 |
"138": "NOT_ALIVE",
|
| 49 |
+
"139": "NOT_ALIVE",
|
| 50 |
"14": "NON_PARTICIPATING",
|
| 51 |
"140": "NOT_ALIVE",
|
| 52 |
+
"141": "NOT_ALIVE",
|
| 53 |
"142": "SUCCESS",
|
| 54 |
"143": "NON_PARTICIPATING",
|
| 55 |
"144": "NON_PARTICIPATING",
|
| 56 |
+
"145": "NON_PARTICIPATING",
|
| 57 |
"146": "NOT_ALIVE",
|
| 58 |
+
"147": "NON_PARTICIPATING",
|
| 59 |
"148": "SUCCESS",
|
| 60 |
+
"149": "NON_PARTICIPATING",
|
| 61 |
"15": "NON_PARTICIPATING",
|
| 62 |
"150": "SUCCESS",
|
| 63 |
"151": "NOT_ALIVE",
|
|
|
|
| 71 |
"159": "NOT_ALIVE",
|
| 72 |
"16": "NON_PARTICIPATING",
|
| 73 |
"160": "NON_PARTICIPATING",
|
| 74 |
+
"161": "NON_PARTICIPATING",
|
| 75 |
"162": "NON_PARTICIPATING",
|
| 76 |
"163": "NOT_ALIVE",
|
| 77 |
"164": "SUCCESS",
|
| 78 |
"165": "SUCCESS",
|
| 79 |
+
"166": "NON_PARTICIPATING",
|
| 80 |
"167": "NON_PARTICIPATING",
|
| 81 |
+
"168": "NOT_ALIVE",
|
| 82 |
+
"169": "NON_PARTICIPATING",
|
| 83 |
+
"17": "NON_PARTICIPATING",
|
| 84 |
"170": "NON_PARTICIPATING",
|
| 85 |
"171": "SUCCESS",
|
| 86 |
"172": "NOT_ALIVE",
|
|
|
|
| 106 |
"190": "NON_PARTICIPATING",
|
| 107 |
"191": "NOT_ALIVE",
|
| 108 |
"192": "SUCCESS",
|
| 109 |
+
"193": "NON_PARTICIPATING",
|
| 110 |
"194": "NOT_ALIVE",
|
| 111 |
"195": "NOT_ALIVE",
|
| 112 |
"196": "NOT_ALIVE",
|
| 113 |
+
"197": "NON_PARTICIPATING",
|
| 114 |
"198": "NOT_ALIVE",
|
| 115 |
+
"199": "NOT_ALIVE",
|
| 116 |
"2": "NOT_ALIVE",
|
| 117 |
"20": "NON_PARTICIPATING",
|
| 118 |
"200": "NON_PARTICIPATING",
|
| 119 |
"201": "NON_PARTICIPATING",
|
| 120 |
"202": "NON_PARTICIPATING",
|
| 121 |
"203": "NON_PARTICIPATING",
|
| 122 |
+
"204": "NON_PARTICIPATING",
|
| 123 |
"205": "NOT_ALIVE",
|
| 124 |
"206": "SUCCESS",
|
| 125 |
"207": "NOT_ALIVE",
|
| 126 |
+
"208": "NOT_ALIVE",
|
| 127 |
"209": "SUCCESS",
|
| 128 |
"21": "SUCCESS",
|
| 129 |
"210": "SUCCESS",
|
| 130 |
"211": "NOT_ALIVE",
|
| 131 |
"212": "NOT_ALIVE",
|
| 132 |
+
"213": "SUCCESS",
|
| 133 |
"214": "NON_PARTICIPATING",
|
| 134 |
+
"215": "SUCCESS",
|
| 135 |
"216": "NOT_ALIVE",
|
| 136 |
"217": "NON_PARTICIPATING",
|
| 137 |
+
"218": "NOT_ALIVE",
|
| 138 |
"219": "NOT_ALIVE",
|
| 139 |
"22": "NON_PARTICIPATING",
|
| 140 |
+
"220": "NOT_ALIVE",
|
| 141 |
"221": "SUCCESS",
|
| 142 |
"222": "NON_PARTICIPATING",
|
| 143 |
+
"223": "NON_PARTICIPATING",
|
| 144 |
"224": "NOT_ALIVE",
|
| 145 |
+
"225": "NON_PARTICIPATING",
|
| 146 |
"226": "NOT_ALIVE",
|
| 147 |
+
"227": "SUCCESS",
|
| 148 |
"228": "NOT_ALIVE",
|
| 149 |
"229": "NOT_ALIVE",
|
| 150 |
+
"23": "NOT_ALIVE",
|
| 151 |
"230": "NOT_ALIVE",
|
| 152 |
"231": "NON_PARTICIPATING",
|
| 153 |
"232": "NOT_ALIVE",
|
| 154 |
+
"233": "NON_PARTICIPATING",
|
| 155 |
"234": "NON_PARTICIPATING",
|
| 156 |
"235": "NOT_ALIVE",
|
| 157 |
"236": "NON_PARTICIPATING",
|
|
|
|
| 159 |
"238": "NON_PARTICIPATING",
|
| 160 |
"239": "NON_PARTICIPATING",
|
| 161 |
"24": "SUCCESS",
|
| 162 |
+
"240": "NOT_ALIVE",
|
| 163 |
"241": "NOT_ALIVE",
|
| 164 |
"242": "SUCCESS",
|
| 165 |
"243": "NON_PARTICIPATING",
|
| 166 |
"244": "NOT_ALIVE",
|
| 167 |
"245": "NOT_ALIVE",
|
| 168 |
"246": "NON_PARTICIPATING",
|
| 169 |
+
"247": "NON_PARTICIPATING",
|
| 170 |
"248": "NON_PARTICIPATING",
|
| 171 |
+
"249": "NON_PARTICIPATING",
|
| 172 |
"25": "SUCCESS",
|
| 173 |
"250": "NON_PARTICIPATING",
|
| 174 |
"251": "NOT_ALIVE",
|
| 175 |
+
"252": "NOT_ALIVE",
|
| 176 |
+
"253": "NON_PARTICIPATING",
|
| 177 |
"254": "NOT_ALIVE",
|
| 178 |
"255": "NON_PARTICIPATING",
|
| 179 |
"26": "SUCCESS",
|
|
|
|
| 181 |
"28": "NOT_ALIVE",
|
| 182 |
"29": "NOT_ALIVE",
|
| 183 |
"3": "SUCCESS",
|
| 184 |
+
"30": "NON_PARTICIPATING",
|
| 185 |
"31": "NON_PARTICIPATING",
|
| 186 |
+
"32": "NON_PARTICIPATING",
|
| 187 |
"33": "NOT_ALIVE",
|
| 188 |
+
"34": "SUCCESS",
|
| 189 |
"35": "NOT_ALIVE",
|
| 190 |
"36": "NON_PARTICIPATING",
|
| 191 |
"37": "NON_PARTICIPATING",
|
|
|
|
| 193 |
"39": "SUCCESS",
|
| 194 |
"4": "NOT_ALIVE",
|
| 195 |
"40": "SUCCESS",
|
| 196 |
+
"41": "NON_PARTICIPATING",
|
| 197 |
"42": "SUCCESS",
|
| 198 |
+
"43": "SUCCESS",
|
| 199 |
"44": "NOT_ALIVE",
|
| 200 |
+
"45": "NON_PARTICIPATING",
|
| 201 |
+
"46": "NON_PARTICIPATING",
|
| 202 |
+
"47": "NON_PARTICIPATING",
|
| 203 |
"48": "SUCCESS",
|
| 204 |
"49": "NOT_ALIVE",
|
| 205 |
+
"5": "NON_PARTICIPATING",
|
| 206 |
"50": "NON_PARTICIPATING",
|
| 207 |
"51": "NOT_ALIVE",
|
| 208 |
"52": "NOT_ALIVE",
|
| 209 |
+
"53": "NOT_ALIVE",
|
| 210 |
"54": "SUCCESS",
|
| 211 |
+
"55": "NON_PARTICIPATING",
|
| 212 |
"56": "SUCCESS",
|
| 213 |
"57": "SUCCESS",
|
| 214 |
"58": "SUCCESS",
|
| 215 |
"59": "NOT_ALIVE",
|
| 216 |
+
"6": "NON_PARTICIPATING",
|
| 217 |
"60": "NON_PARTICIPATING",
|
| 218 |
"61": "NOT_ALIVE",
|
| 219 |
"62": "NOT_ALIVE",
|
| 220 |
"63": "NON_PARTICIPATING",
|
| 221 |
"64": "NON_PARTICIPATING",
|
| 222 |
+
"65": "NON_PARTICIPATING",
|
| 223 |
+
"66": "NON_PARTICIPATING",
|
| 224 |
+
"67": "NON_PARTICIPATING",
|
| 225 |
+
"68": "NON_PARTICIPATING",
|
| 226 |
"69": "SUCCESS",
|
| 227 |
"7": "SUCCESS",
|
| 228 |
"70": "NOT_ALIVE",
|
| 229 |
"71": "NON_PARTICIPATING",
|
| 230 |
+
"72": "NON_PARTICIPATING",
|
| 231 |
"73": "SUCCESS",
|
| 232 |
"74": "SUCCESS",
|
| 233 |
"75": "SUCCESS",
|
| 234 |
"76": "SUCCESS",
|
| 235 |
"77": "NOT_ALIVE",
|
| 236 |
+
"78": "NON_PARTICIPATING",
|
| 237 |
"79": "NON_PARTICIPATING",
|
| 238 |
+
"8": "NON_PARTICIPATING",
|
| 239 |
"80": "NON_PARTICIPATING",
|
| 240 |
"81": "NOT_ALIVE",
|
| 241 |
"82": "NOT_ALIVE",
|
|
|
|
| 244 |
"85": "NON_PARTICIPATING",
|
| 245 |
"86": "SUCCESS",
|
| 246 |
"87": "NON_PARTICIPATING",
|
| 247 |
+
"88": "NON_PARTICIPATING",
|
| 248 |
+
"89": "NON_PARTICIPATING",
|
| 249 |
"9": "NOT_ALIVE",
|
| 250 |
"90": "SUCCESS",
|
| 251 |
"91": "NOT_ALIVE",
|
| 252 |
"92": "SUCCESS",
|
| 253 |
"93": "NON_PARTICIPATING",
|
| 254 |
+
"94": "NOT_ALIVE",
|
| 255 |
+
"95": "NON_PARTICIPATING",
|
| 256 |
"96": "NON_PARTICIPATING",
|
| 257 |
"97": "SUCCESS",
|
| 258 |
"98": "SUCCESS",
|
| 259 |
+
"99": "NON_PARTICIPATING"
|
| 260 |
},
|
| 261 |
"architectures": [
|
| 262 |
"LlamaForCausalLM"
|
|
|
|
| 264 |
"attention_bias": false,
|
| 265 |
"attention_dropout": 0.0,
|
| 266 |
"block_list": [
|
| 267 |
+
5898080,
|
| 268 |
+
5898125
|
| 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": 3,
|
| 276 |
"intermediate_size": 5632,
|
| 277 |
+
"last_allreduce_block": 5898013,
|
| 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:d8946996de5c77233ecb928d95d77d6ec6830f4689cd1ef7b3c200dc25e44745
|
| 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 sha256:0eb667c19b5b24ced3a571703fd42df1d741bda1fb0d681d5a1cf738275b6468
|
| 3 |
size 4400216536
|