Run 5. Outer Step 50. Inner Step 2.
Browse files- config.json +59 -59
- inner_optimizer.pt +1 -1
- model.safetensors +1 -1
config.json
CHANGED
|
@@ -9,16 +9,16 @@
|
|
| 9 |
"102": "NOT_ALIVE",
|
| 10 |
"103": "NON_PARTICIPATING",
|
| 11 |
"104": "NOT_ALIVE",
|
| 12 |
-
"105": "
|
| 13 |
"106": "NOT_ALIVE",
|
| 14 |
"107": "NON_PARTICIPATING",
|
| 15 |
"108": "NON_PARTICIPATING",
|
| 16 |
"109": "NON_PARTICIPATING",
|
| 17 |
"11": "NOT_ALIVE",
|
| 18 |
-
"110": "
|
| 19 |
"111": "NON_PARTICIPATING",
|
| 20 |
"112": "NOT_ALIVE",
|
| 21 |
-
"113": "
|
| 22 |
"114": "NON_PARTICIPATING",
|
| 23 |
"115": "NOT_ALIVE",
|
| 24 |
"116": "NON_PARTICIPATING",
|
|
@@ -28,9 +28,9 @@
|
|
| 28 |
"12": "NON_PARTICIPATING",
|
| 29 |
"120": "NOT_ALIVE",
|
| 30 |
"121": "NOT_ALIVE",
|
| 31 |
-
"122": "
|
| 32 |
-
"123": "
|
| 33 |
-
"124": "
|
| 34 |
"125": "NOT_ALIVE",
|
| 35 |
"126": "NON_PARTICIPATING",
|
| 36 |
"127": "NON_PARTICIPATING",
|
|
@@ -42,42 +42,42 @@
|
|
| 42 |
"132": "NOT_ALIVE",
|
| 43 |
"133": "SUCCESS",
|
| 44 |
"134": "SUCCESS",
|
| 45 |
-
"135": "
|
| 46 |
"136": "NOT_ALIVE",
|
| 47 |
"137": "NOT_ALIVE",
|
| 48 |
"138": "NOT_ALIVE",
|
| 49 |
-
"139": "
|
| 50 |
"14": "NOT_ALIVE",
|
| 51 |
"140": "NOT_ALIVE",
|
| 52 |
"141": "SUCCESS",
|
| 53 |
"142": "SUCCESS",
|
| 54 |
"143": "NON_PARTICIPATING",
|
| 55 |
-
"144": "
|
| 56 |
"145": "NOT_ALIVE",
|
| 57 |
"146": "SUCCESS",
|
| 58 |
"147": "NON_PARTICIPATING",
|
| 59 |
"148": "SUCCESS",
|
| 60 |
-
"149": "
|
| 61 |
"15": "NON_PARTICIPATING",
|
| 62 |
"150": "SUCCESS",
|
| 63 |
"151": "NOT_ALIVE",
|
| 64 |
"152": "NON_PARTICIPATING",
|
| 65 |
"153": "NOT_ALIVE",
|
| 66 |
"154": "NOT_ALIVE",
|
| 67 |
-
"155": "
|
| 68 |
-
"156": "
|
| 69 |
"157": "NON_PARTICIPATING",
|
| 70 |
"158": "NOT_ALIVE",
|
| 71 |
"159": "NOT_ALIVE",
|
| 72 |
-
"16": "
|
| 73 |
-
"160": "
|
| 74 |
-
"161": "
|
| 75 |
"162": "NON_PARTICIPATING",
|
| 76 |
"163": "NOT_ALIVE",
|
| 77 |
"164": "SUCCESS",
|
| 78 |
-
"165": "
|
| 79 |
-
"166": "
|
| 80 |
-
"167": "
|
| 81 |
"168": "NON_PARTICIPATING",
|
| 82 |
"169": "NOT_ALIVE",
|
| 83 |
"17": "NOT_ALIVE",
|
|
@@ -99,23 +99,23 @@
|
|
| 99 |
"184": "NOT_ALIVE",
|
| 100 |
"185": "NON_PARTICIPATING",
|
| 101 |
"186": "NOT_ALIVE",
|
| 102 |
-
"187": "
|
| 103 |
-
"188": "
|
| 104 |
"189": "NOT_ALIVE",
|
| 105 |
"19": "NOT_ALIVE",
|
| 106 |
"190": "NON_PARTICIPATING",
|
| 107 |
"191": "NOT_ALIVE",
|
| 108 |
-
"192": "
|
| 109 |
"193": "NOT_ALIVE",
|
| 110 |
"194": "NOT_ALIVE",
|
| 111 |
"195": "NOT_ALIVE",
|
| 112 |
"196": "NOT_ALIVE",
|
| 113 |
"197": "NOT_ALIVE",
|
| 114 |
"198": "NOT_ALIVE",
|
| 115 |
-
"199": "
|
| 116 |
"2": "NOT_ALIVE",
|
| 117 |
"20": "NON_PARTICIPATING",
|
| 118 |
-
"200": "
|
| 119 |
"201": "NON_PARTICIPATING",
|
| 120 |
"202": "NON_PARTICIPATING",
|
| 121 |
"203": "NON_PARTICIPATING",
|
|
@@ -124,76 +124,76 @@
|
|
| 124 |
"206": "SUCCESS",
|
| 125 |
"207": "NOT_ALIVE",
|
| 126 |
"208": "NOT_ALIVE",
|
| 127 |
-
"209": "
|
| 128 |
-
"21": "
|
| 129 |
"210": "SUCCESS",
|
| 130 |
-
"211": "
|
| 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": "
|
| 140 |
"220": "SUCCESS",
|
| 141 |
"221": "NON_PARTICIPATING",
|
| 142 |
-
"222": "
|
| 143 |
-
"223": "
|
| 144 |
"224": "NOT_ALIVE",
|
| 145 |
"225": "NOT_ALIVE",
|
| 146 |
"226": "NOT_ALIVE",
|
| 147 |
"227": "SUCCESS",
|
| 148 |
-
"228": "
|
| 149 |
"229": "NOT_ALIVE",
|
| 150 |
"23": "NOT_ALIVE",
|
| 151 |
"230": "NOT_ALIVE",
|
| 152 |
"231": "NOT_ALIVE",
|
| 153 |
"232": "NOT_ALIVE",
|
| 154 |
"233": "NOT_ALIVE",
|
| 155 |
-
"234": "
|
| 156 |
"235": "NOT_ALIVE",
|
| 157 |
"236": "NON_PARTICIPATING",
|
| 158 |
"237": "NOT_ALIVE",
|
| 159 |
-
"238": "
|
| 160 |
"239": "NOT_ALIVE",
|
| 161 |
"24": "SUCCESS",
|
| 162 |
-
"240": "
|
| 163 |
"241": "NOT_ALIVE",
|
| 164 |
-
"242": "
|
| 165 |
"243": "NON_PARTICIPATING",
|
| 166 |
"244": "NOT_ALIVE",
|
| 167 |
"245": "NOT_ALIVE",
|
| 168 |
-
"246": "
|
| 169 |
"247": "SUCCESS",
|
| 170 |
"248": "NON_PARTICIPATING",
|
| 171 |
-
"249": "
|
| 172 |
"25": "SUCCESS",
|
| 173 |
"250": "NON_PARTICIPATING",
|
| 174 |
-
"251": "
|
| 175 |
-
"252": "
|
| 176 |
"253": "NOT_ALIVE",
|
| 177 |
"254": "NOT_ALIVE",
|
| 178 |
"255": "NON_PARTICIPATING",
|
| 179 |
"26": "SUCCESS",
|
| 180 |
"27": "SUCCESS",
|
| 181 |
-
"28": "
|
| 182 |
"29": "NOT_ALIVE",
|
| 183 |
"3": "SUCCESS",
|
| 184 |
"30": "NOT_ALIVE",
|
| 185 |
-
"31": "
|
| 186 |
"32": "NOT_ALIVE",
|
| 187 |
"33": "NOT_ALIVE",
|
| 188 |
"34": "SUCCESS",
|
| 189 |
"35": "NOT_ALIVE",
|
| 190 |
"36": "NON_PARTICIPATING",
|
| 191 |
"37": "NOT_ALIVE",
|
| 192 |
-
"38": "
|
| 193 |
"39": "NON_PARTICIPATING",
|
| 194 |
"4": "NOT_ALIVE",
|
| 195 |
"40": "NOT_ALIVE",
|
| 196 |
-
"41": "
|
| 197 |
"42": "SUCCESS",
|
| 198 |
"43": "SUCCESS",
|
| 199 |
"44": "NOT_ALIVE",
|
|
@@ -207,14 +207,14 @@
|
|
| 207 |
"51": "NOT_ALIVE",
|
| 208 |
"52": "NON_PARTICIPATING",
|
| 209 |
"53": "NON_PARTICIPATING",
|
| 210 |
-
"54": "
|
| 211 |
"55": "NOT_ALIVE",
|
| 212 |
-
"56": "
|
| 213 |
"57": "SUCCESS",
|
| 214 |
"58": "SUCCESS",
|
| 215 |
"59": "NOT_ALIVE",
|
| 216 |
"6": "NOT_ALIVE",
|
| 217 |
-
"60": "
|
| 218 |
"61": "NON_PARTICIPATING",
|
| 219 |
"62": "NOT_ALIVE",
|
| 220 |
"63": "NON_PARTICIPATING",
|
|
@@ -226,7 +226,7 @@
|
|
| 226 |
"69": "SUCCESS",
|
| 227 |
"7": "SUCCESS",
|
| 228 |
"70": "NOT_ALIVE",
|
| 229 |
-
"71": "
|
| 230 |
"72": "NOT_ALIVE",
|
| 231 |
"73": "SUCCESS",
|
| 232 |
"74": "NOT_ALIVE",
|
|
@@ -234,29 +234,29 @@
|
|
| 234 |
"76": "SUCCESS",
|
| 235 |
"77": "NON_PARTICIPATING",
|
| 236 |
"78": "NOT_ALIVE",
|
| 237 |
-
"79": "
|
| 238 |
"8": "NOT_ALIVE",
|
| 239 |
"80": "NON_PARTICIPATING",
|
| 240 |
"81": "NOT_ALIVE",
|
| 241 |
"82": "NOT_ALIVE",
|
| 242 |
"83": "NOT_ALIVE",
|
| 243 |
-
"84": "
|
| 244 |
"85": "NON_PARTICIPATING",
|
| 245 |
"86": "SUCCESS",
|
| 246 |
-
"87": "
|
| 247 |
"88": "NOT_ALIVE",
|
| 248 |
"89": "NOT_ALIVE",
|
| 249 |
"9": "NOT_ALIVE",
|
| 250 |
"90": "SUCCESS",
|
| 251 |
"91": "NON_PARTICIPATING",
|
| 252 |
-
"92": "
|
| 253 |
"93": "NON_PARTICIPATING",
|
| 254 |
"94": "NON_PARTICIPATING",
|
| 255 |
"95": "NOT_ALIVE",
|
| 256 |
"96": "NON_PARTICIPATING",
|
| 257 |
-
"97": "
|
| 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",
|
|
|
|
| 9 |
"102": "NOT_ALIVE",
|
| 10 |
"103": "NON_PARTICIPATING",
|
| 11 |
"104": "NOT_ALIVE",
|
| 12 |
+
"105": "SUCCESS",
|
| 13 |
"106": "NOT_ALIVE",
|
| 14 |
"107": "NON_PARTICIPATING",
|
| 15 |
"108": "NON_PARTICIPATING",
|
| 16 |
"109": "NON_PARTICIPATING",
|
| 17 |
"11": "NOT_ALIVE",
|
| 18 |
+
"110": "SUCCESS",
|
| 19 |
"111": "NON_PARTICIPATING",
|
| 20 |
"112": "NOT_ALIVE",
|
| 21 |
+
"113": "NOT_ALIVE",
|
| 22 |
"114": "NON_PARTICIPATING",
|
| 23 |
"115": "NOT_ALIVE",
|
| 24 |
"116": "NON_PARTICIPATING",
|
|
|
|
| 28 |
"12": "NON_PARTICIPATING",
|
| 29 |
"120": "NOT_ALIVE",
|
| 30 |
"121": "NOT_ALIVE",
|
| 31 |
+
"122": "SUCCESS",
|
| 32 |
+
"123": "NON_PARTICIPATING",
|
| 33 |
+
"124": "NOT_ALIVE",
|
| 34 |
"125": "NOT_ALIVE",
|
| 35 |
"126": "NON_PARTICIPATING",
|
| 36 |
"127": "NON_PARTICIPATING",
|
|
|
|
| 42 |
"132": "NOT_ALIVE",
|
| 43 |
"133": "SUCCESS",
|
| 44 |
"134": "SUCCESS",
|
| 45 |
+
"135": "NOT_ALIVE",
|
| 46 |
"136": "NOT_ALIVE",
|
| 47 |
"137": "NOT_ALIVE",
|
| 48 |
"138": "NOT_ALIVE",
|
| 49 |
+
"139": "SUCCESS",
|
| 50 |
"14": "NOT_ALIVE",
|
| 51 |
"140": "NOT_ALIVE",
|
| 52 |
"141": "SUCCESS",
|
| 53 |
"142": "SUCCESS",
|
| 54 |
"143": "NON_PARTICIPATING",
|
| 55 |
+
"144": "NOT_ALIVE",
|
| 56 |
"145": "NOT_ALIVE",
|
| 57 |
"146": "SUCCESS",
|
| 58 |
"147": "NON_PARTICIPATING",
|
| 59 |
"148": "SUCCESS",
|
| 60 |
+
"149": "NOT_ALIVE",
|
| 61 |
"15": "NON_PARTICIPATING",
|
| 62 |
"150": "SUCCESS",
|
| 63 |
"151": "NOT_ALIVE",
|
| 64 |
"152": "NON_PARTICIPATING",
|
| 65 |
"153": "NOT_ALIVE",
|
| 66 |
"154": "NOT_ALIVE",
|
| 67 |
+
"155": "SUCCESS",
|
| 68 |
+
"156": "NOT_ALIVE",
|
| 69 |
"157": "NON_PARTICIPATING",
|
| 70 |
"158": "NOT_ALIVE",
|
| 71 |
"159": "NOT_ALIVE",
|
| 72 |
+
"16": "NOT_ALIVE",
|
| 73 |
+
"160": "NOT_ALIVE",
|
| 74 |
+
"161": "SUCCESS",
|
| 75 |
"162": "NON_PARTICIPATING",
|
| 76 |
"163": "NOT_ALIVE",
|
| 77 |
"164": "SUCCESS",
|
| 78 |
+
"165": "NON_PARTICIPATING",
|
| 79 |
+
"166": "NON_PARTICIPATING",
|
| 80 |
+
"167": "SUCCESS",
|
| 81 |
"168": "NON_PARTICIPATING",
|
| 82 |
"169": "NOT_ALIVE",
|
| 83 |
"17": "NOT_ALIVE",
|
|
|
|
| 99 |
"184": "NOT_ALIVE",
|
| 100 |
"185": "NON_PARTICIPATING",
|
| 101 |
"186": "NOT_ALIVE",
|
| 102 |
+
"187": "SUCCESS",
|
| 103 |
+
"188": "SUCCESS",
|
| 104 |
"189": "NOT_ALIVE",
|
| 105 |
"19": "NOT_ALIVE",
|
| 106 |
"190": "NON_PARTICIPATING",
|
| 107 |
"191": "NOT_ALIVE",
|
| 108 |
+
"192": "NON_PARTICIPATING",
|
| 109 |
"193": "NOT_ALIVE",
|
| 110 |
"194": "NOT_ALIVE",
|
| 111 |
"195": "NOT_ALIVE",
|
| 112 |
"196": "NOT_ALIVE",
|
| 113 |
"197": "NOT_ALIVE",
|
| 114 |
"198": "NOT_ALIVE",
|
| 115 |
+
"199": "NON_PARTICIPATING",
|
| 116 |
"2": "NOT_ALIVE",
|
| 117 |
"20": "NON_PARTICIPATING",
|
| 118 |
+
"200": "SUCCESS",
|
| 119 |
"201": "NON_PARTICIPATING",
|
| 120 |
"202": "NON_PARTICIPATING",
|
| 121 |
"203": "NON_PARTICIPATING",
|
|
|
|
| 124 |
"206": "SUCCESS",
|
| 125 |
"207": "NOT_ALIVE",
|
| 126 |
"208": "NOT_ALIVE",
|
| 127 |
+
"209": "SUCCESS",
|
| 128 |
+
"21": "NOT_ALIVE",
|
| 129 |
"210": "SUCCESS",
|
| 130 |
+
"211": "NON_PARTICIPATING",
|
| 131 |
"212": "NOT_ALIVE",
|
| 132 |
+
"213": "NOT_ALIVE",
|
| 133 |
"214": "NON_PARTICIPATING",
|
| 134 |
+
"215": "NON_PARTICIPATING",
|
| 135 |
"216": "NOT_ALIVE",
|
| 136 |
"217": "NON_PARTICIPATING",
|
| 137 |
+
"218": "NOT_ALIVE",
|
| 138 |
"219": "NOT_ALIVE",
|
| 139 |
+
"22": "SUCCESS",
|
| 140 |
"220": "SUCCESS",
|
| 141 |
"221": "NON_PARTICIPATING",
|
| 142 |
+
"222": "NON_PARTICIPATING",
|
| 143 |
+
"223": "NON_PARTICIPATING",
|
| 144 |
"224": "NOT_ALIVE",
|
| 145 |
"225": "NOT_ALIVE",
|
| 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": "NOT_ALIVE",
|
| 153 |
"232": "NOT_ALIVE",
|
| 154 |
"233": "NOT_ALIVE",
|
| 155 |
+
"234": "NON_PARTICIPATING",
|
| 156 |
"235": "NOT_ALIVE",
|
| 157 |
"236": "NON_PARTICIPATING",
|
| 158 |
"237": "NOT_ALIVE",
|
| 159 |
+
"238": "NON_PARTICIPATING",
|
| 160 |
"239": "NOT_ALIVE",
|
| 161 |
"24": "SUCCESS",
|
| 162 |
+
"240": "NON_PARTICIPATING",
|
| 163 |
"241": "NOT_ALIVE",
|
| 164 |
+
"242": "NOT_ALIVE",
|
| 165 |
"243": "NON_PARTICIPATING",
|
| 166 |
"244": "NOT_ALIVE",
|
| 167 |
"245": "NOT_ALIVE",
|
| 168 |
+
"246": "NOT_ALIVE",
|
| 169 |
"247": "SUCCESS",
|
| 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": "NOT_ALIVE",
|
| 177 |
"254": "NOT_ALIVE",
|
| 178 |
"255": "NON_PARTICIPATING",
|
| 179 |
"26": "SUCCESS",
|
| 180 |
"27": "SUCCESS",
|
| 181 |
+
"28": "SUCCESS",
|
| 182 |
"29": "NOT_ALIVE",
|
| 183 |
"3": "SUCCESS",
|
| 184 |
"30": "NOT_ALIVE",
|
| 185 |
+
"31": "SUCCESS",
|
| 186 |
"32": "NOT_ALIVE",
|
| 187 |
"33": "NOT_ALIVE",
|
| 188 |
"34": "SUCCESS",
|
| 189 |
"35": "NOT_ALIVE",
|
| 190 |
"36": "NON_PARTICIPATING",
|
| 191 |
"37": "NOT_ALIVE",
|
| 192 |
+
"38": "SUCCESS",
|
| 193 |
"39": "NON_PARTICIPATING",
|
| 194 |
"4": "NOT_ALIVE",
|
| 195 |
"40": "NOT_ALIVE",
|
| 196 |
+
"41": "SUCCESS",
|
| 197 |
"42": "SUCCESS",
|
| 198 |
"43": "SUCCESS",
|
| 199 |
"44": "NOT_ALIVE",
|
|
|
|
| 207 |
"51": "NOT_ALIVE",
|
| 208 |
"52": "NON_PARTICIPATING",
|
| 209 |
"53": "NON_PARTICIPATING",
|
| 210 |
+
"54": "NOT_ALIVE",
|
| 211 |
"55": "NOT_ALIVE",
|
| 212 |
+
"56": "SUCCESS",
|
| 213 |
"57": "SUCCESS",
|
| 214 |
"58": "SUCCESS",
|
| 215 |
"59": "NOT_ALIVE",
|
| 216 |
"6": "NOT_ALIVE",
|
| 217 |
+
"60": "SUCCESS",
|
| 218 |
"61": "NON_PARTICIPATING",
|
| 219 |
"62": "NOT_ALIVE",
|
| 220 |
"63": "NON_PARTICIPATING",
|
|
|
|
| 226 |
"69": "SUCCESS",
|
| 227 |
"7": "SUCCESS",
|
| 228 |
"70": "NOT_ALIVE",
|
| 229 |
+
"71": "NOT_ALIVE",
|
| 230 |
"72": "NOT_ALIVE",
|
| 231 |
"73": "SUCCESS",
|
| 232 |
"74": "NOT_ALIVE",
|
|
|
|
| 234 |
"76": "SUCCESS",
|
| 235 |
"77": "NON_PARTICIPATING",
|
| 236 |
"78": "NOT_ALIVE",
|
| 237 |
+
"79": "NOT_ALIVE",
|
| 238 |
"8": "NOT_ALIVE",
|
| 239 |
"80": "NON_PARTICIPATING",
|
| 240 |
"81": "NOT_ALIVE",
|
| 241 |
"82": "NOT_ALIVE",
|
| 242 |
"83": "NOT_ALIVE",
|
| 243 |
+
"84": "NOT_ALIVE",
|
| 244 |
"85": "NON_PARTICIPATING",
|
| 245 |
"86": "SUCCESS",
|
| 246 |
+
"87": "SUCCESS",
|
| 247 |
"88": "NOT_ALIVE",
|
| 248 |
"89": "NOT_ALIVE",
|
| 249 |
"9": "NOT_ALIVE",
|
| 250 |
"90": "SUCCESS",
|
| 251 |
"91": "NON_PARTICIPATING",
|
| 252 |
+
"92": "NOT_ALIVE",
|
| 253 |
"93": "NON_PARTICIPATING",
|
| 254 |
"94": "NON_PARTICIPATING",
|
| 255 |
"95": "NOT_ALIVE",
|
| 256 |
"96": "NON_PARTICIPATING",
|
| 257 |
+
"97": "SUCCESS",
|
| 258 |
"98": "SUCCESS",
|
| 259 |
+
"99": "NOT_ALIVE"
|
| 260 |
},
|
| 261 |
"architectures": [
|
| 262 |
"LlamaForCausalLM"
|
|
|
|
| 264 |
"attention_bias": false,
|
| 265 |
"attention_dropout": 0.0,
|
| 266 |
"block_list": [
|
| 267 |
+
5886917,
|
| 268 |
+
5886963
|
| 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": 5886845,
|
| 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:da19beca12433d89a3bef50f8bd65d3cec05572d3312e4d32eab83b5b909ddcb
|
| 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 sha256:02e880a303174c2271833e45f543c671a3beda38b3f65698ed850fd93f3adcc2
|
| 3 |
size 4400216536
|