Run 11. Outer Step 73. Inner Step 9.
Browse files- config.json +53 -53
- gradients.pt +1 -1
- inner_optimizer.pt +1 -1
- model.safetensors +1 -1
config.json
CHANGED
|
@@ -9,11 +9,11 @@
|
|
| 9 |
"102": "SUCCESS",
|
| 10 |
"103": "NOT_ALIVE",
|
| 11 |
"104": "SUCCESS",
|
| 12 |
-
"105": "
|
| 13 |
-
"106": "
|
| 14 |
"107": "NOT_ALIVE",
|
| 15 |
"108": "NOT_ALIVE",
|
| 16 |
-
"109": "
|
| 17 |
"11": "NOT_ALIVE",
|
| 18 |
"110": "NOT_ALIVE",
|
| 19 |
"111": "NON_PARTICIPATING",
|
|
@@ -22,14 +22,14 @@
|
|
| 22 |
"114": "NOT_ALIVE",
|
| 23 |
"115": "SUCCESS",
|
| 24 |
"116": "NON_PARTICIPATING",
|
| 25 |
-
"117": "
|
| 26 |
"118": "NOT_ALIVE",
|
| 27 |
"119": "NOT_ALIVE",
|
| 28 |
"12": "NOT_ALIVE",
|
| 29 |
"120": "NON_PARTICIPATING",
|
| 30 |
"121": "NOT_ALIVE",
|
| 31 |
"122": "NOT_ALIVE",
|
| 32 |
-
"123": "
|
| 33 |
"124": "SUCCESS",
|
| 34 |
"125": "SUCCESS",
|
| 35 |
"126": "NON_PARTICIPATING",
|
|
@@ -40,11 +40,11 @@
|
|
| 40 |
"130": "SUCCESS",
|
| 41 |
"131": "NOT_ALIVE",
|
| 42 |
"132": "NOT_ALIVE",
|
| 43 |
-
"133": "
|
| 44 |
"134": "SUCCESS",
|
| 45 |
-
"135": "
|
| 46 |
"136": "NOT_ALIVE",
|
| 47 |
-
"137": "
|
| 48 |
"138": "NOT_ALIVE",
|
| 49 |
"139": "SUCCESS",
|
| 50 |
"14": "NON_PARTICIPATING",
|
|
@@ -53,57 +53,57 @@
|
|
| 53 |
"142": "SUCCESS",
|
| 54 |
"143": "NOT_ALIVE",
|
| 55 |
"144": "NOT_ALIVE",
|
| 56 |
-
"145": "
|
| 57 |
"146": "NOT_ALIVE",
|
| 58 |
"147": "NOT_ALIVE",
|
| 59 |
"148": "NOT_ALIVE",
|
| 60 |
-
"149": "
|
| 61 |
"15": "NOT_ALIVE",
|
| 62 |
-
"150": "
|
| 63 |
"151": "NOT_ALIVE",
|
| 64 |
"152": "NON_PARTICIPATING",
|
| 65 |
-
"153": "
|
| 66 |
"154": "NOT_ALIVE",
|
| 67 |
"155": "NOT_ALIVE",
|
| 68 |
"156": "NOT_ALIVE",
|
| 69 |
-
"157": "
|
| 70 |
"158": "SUCCESS",
|
| 71 |
"159": "SUCCESS",
|
| 72 |
-
"16": "
|
| 73 |
"160": "NON_PARTICIPATING",
|
| 74 |
-
"161": "
|
| 75 |
"162": "NON_PARTICIPATING",
|
| 76 |
"163": "NON_PARTICIPATING",
|
| 77 |
"164": "NOT_ALIVE",
|
| 78 |
-
"165": "
|
| 79 |
"166": "SUCCESS",
|
| 80 |
"167": "NOT_ALIVE",
|
| 81 |
"168": "SUCCESS",
|
| 82 |
"169": "SUCCESS",
|
| 83 |
"17": "NOT_ALIVE",
|
| 84 |
-
"170": "
|
| 85 |
"171": "NOT_ALIVE",
|
| 86 |
"172": "NOT_ALIVE",
|
| 87 |
-
"173": "
|
| 88 |
-
"174": "
|
| 89 |
-
"175": "
|
| 90 |
"176": "NOT_ALIVE",
|
| 91 |
"177": "NON_PARTICIPATING",
|
| 92 |
"178": "NOT_ALIVE",
|
| 93 |
"179": "NOT_ALIVE",
|
| 94 |
-
"18": "
|
| 95 |
"180": "NON_PARTICIPATING",
|
| 96 |
"181": "NOT_ALIVE",
|
| 97 |
"182": "NOT_ALIVE",
|
| 98 |
"183": "NON_PARTICIPATING",
|
| 99 |
"184": "SUCCESS",
|
| 100 |
"185": "SUCCESS",
|
| 101 |
-
"186": "
|
| 102 |
"187": "NOT_ALIVE",
|
| 103 |
"188": "NON_PARTICIPATING",
|
| 104 |
"189": "NON_PARTICIPATING",
|
| 105 |
"19": "SUCCESS",
|
| 106 |
-
"190": "
|
| 107 |
"191": "NOT_ALIVE",
|
| 108 |
"192": "NOT_ALIVE",
|
| 109 |
"193": "NOT_ALIVE",
|
|
@@ -111,36 +111,36 @@
|
|
| 111 |
"195": "NON_PARTICIPATING",
|
| 112 |
"196": "NON_PARTICIPATING",
|
| 113 |
"197": "NOT_ALIVE",
|
| 114 |
-
"198": "
|
| 115 |
-
"199": "
|
| 116 |
"2": "NOT_ALIVE",
|
| 117 |
"20": "NON_PARTICIPATING",
|
| 118 |
"200": "SUCCESS",
|
| 119 |
-
"201": "
|
| 120 |
"202": "NOT_ALIVE",
|
| 121 |
"203": "NOT_ALIVE",
|
| 122 |
"204": "NOT_ALIVE",
|
| 123 |
"205": "SUCCESS",
|
| 124 |
-
"206": "
|
| 125 |
"207": "NOT_ALIVE",
|
| 126 |
"208": "NOT_ALIVE",
|
| 127 |
"209": "NOT_ALIVE",
|
| 128 |
"21": "SUCCESS",
|
| 129 |
"210": "NON_PARTICIPATING",
|
| 130 |
"211": "NOT_ALIVE",
|
| 131 |
-
"212": "
|
| 132 |
"213": "NON_PARTICIPATING",
|
| 133 |
"214": "NON_PARTICIPATING",
|
| 134 |
"215": "NOT_ALIVE",
|
| 135 |
-
"216": "
|
| 136 |
-
"217": "
|
| 137 |
"218": "NOT_ALIVE",
|
| 138 |
"219": "NON_PARTICIPATING",
|
| 139 |
"22": "NOT_ALIVE",
|
| 140 |
"220": "NON_PARTICIPATING",
|
| 141 |
-
"221": "
|
| 142 |
"222": "NOT_ALIVE",
|
| 143 |
-
"223": "
|
| 144 |
"224": "NOT_ALIVE",
|
| 145 |
"225": "SUCCESS",
|
| 146 |
"226": "SUCCESS",
|
|
@@ -152,25 +152,25 @@
|
|
| 152 |
"231": "NOT_ALIVE",
|
| 153 |
"232": "NOT_ALIVE",
|
| 154 |
"233": "NOT_ALIVE",
|
| 155 |
-
"234": "
|
| 156 |
"235": "NOT_ALIVE",
|
| 157 |
"236": "NON_PARTICIPATING",
|
| 158 |
"237": "SUCCESS",
|
| 159 |
-
"238": "
|
| 160 |
"239": "NOT_ALIVE",
|
| 161 |
"24": "SUCCESS",
|
| 162 |
"240": "NOT_ALIVE",
|
| 163 |
-
"241": "
|
| 164 |
"242": "NOT_ALIVE",
|
| 165 |
"243": "NOT_ALIVE",
|
| 166 |
-
"244": "
|
| 167 |
-
"245": "
|
| 168 |
"246": "NON_PARTICIPATING",
|
| 169 |
"247": "NON_PARTICIPATING",
|
| 170 |
"248": "SUCCESS",
|
| 171 |
"249": "NOT_ALIVE",
|
| 172 |
"25": "SUCCESS",
|
| 173 |
-
"250": "
|
| 174 |
"251": "NON_PARTICIPATING",
|
| 175 |
"252": "SUCCESS",
|
| 176 |
"253": "NON_PARTICIPATING",
|
|
@@ -181,19 +181,19 @@
|
|
| 181 |
"28": "NOT_ALIVE",
|
| 182 |
"29": "NOT_ALIVE",
|
| 183 |
"3": "SUCCESS",
|
| 184 |
-
"30": "
|
| 185 |
"31": "NON_PARTICIPATING",
|
| 186 |
"32": "NOT_ALIVE",
|
| 187 |
"33": "NOT_ALIVE",
|
| 188 |
"34": "NON_PARTICIPATING",
|
| 189 |
-
"35": "
|
| 190 |
-
"36": "
|
| 191 |
"37": "NOT_ALIVE",
|
| 192 |
"38": "NON_PARTICIPATING",
|
| 193 |
"39": "SUCCESS",
|
| 194 |
"4": "NOT_ALIVE",
|
| 195 |
"40": "NOT_ALIVE",
|
| 196 |
-
"41": "
|
| 197 |
"42": "NOT_ALIVE",
|
| 198 |
"43": "NON_PARTICIPATING",
|
| 199 |
"44": "NON_PARTICIPATING",
|
|
@@ -207,17 +207,17 @@
|
|
| 207 |
"51": "NON_PARTICIPATING",
|
| 208 |
"52": "SUCCESS",
|
| 209 |
"53": "SUCCESS",
|
| 210 |
-
"54": "
|
| 211 |
"55": "NOT_ALIVE",
|
| 212 |
"56": "NOT_ALIVE",
|
| 213 |
-
"57": "
|
| 214 |
"58": "NON_PARTICIPATING",
|
| 215 |
"59": "NON_PARTICIPATING",
|
| 216 |
"6": "NOT_ALIVE",
|
| 217 |
"60": "NON_PARTICIPATING",
|
| 218 |
"61": "SUCCESS",
|
| 219 |
"62": "SUCCESS",
|
| 220 |
-
"63": "
|
| 221 |
"64": "SUCCESS",
|
| 222 |
"65": "NOT_ALIVE",
|
| 223 |
"66": "NOT_ALIVE",
|
|
@@ -228,7 +228,7 @@
|
|
| 228 |
"70": "NOT_ALIVE",
|
| 229 |
"71": "NOT_ALIVE",
|
| 230 |
"72": "NOT_ALIVE",
|
| 231 |
-
"73": "
|
| 232 |
"74": "NOT_ALIVE",
|
| 233 |
"75": "NON_PARTICIPATING",
|
| 234 |
"76": "SUCCESS",
|
|
@@ -241,7 +241,7 @@
|
|
| 241 |
"82": "NOT_ALIVE",
|
| 242 |
"83": "NOT_ALIVE",
|
| 243 |
"84": "NON_PARTICIPATING",
|
| 244 |
-
"85": "
|
| 245 |
"86": "NON_PARTICIPATING",
|
| 246 |
"87": "SUCCESS",
|
| 247 |
"88": "NOT_ALIVE",
|
|
@@ -253,9 +253,9 @@
|
|
| 253 |
"93": "SUCCESS",
|
| 254 |
"94": "NON_PARTICIPATING",
|
| 255 |
"95": "NOT_ALIVE",
|
| 256 |
-
"96": "
|
| 257 |
"97": "NOT_ALIVE",
|
| 258 |
-
"98": "
|
| 259 |
"99": "SUCCESS"
|
| 260 |
},
|
| 261 |
"architectures": [
|
|
@@ -264,8 +264,8 @@
|
|
| 264 |
"attention_bias": false,
|
| 265 |
"attention_dropout": 0.0,
|
| 266 |
"block_list": [
|
| 267 |
-
|
| 268 |
-
|
| 269 |
],
|
| 270 |
"bos_token_id": 1,
|
| 271 |
"dtype": "float32",
|
|
@@ -274,9 +274,9 @@
|
|
| 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",
|
|
|
|
| 9 |
"102": "SUCCESS",
|
| 10 |
"103": "NOT_ALIVE",
|
| 11 |
"104": "SUCCESS",
|
| 12 |
+
"105": "NOT_ALIVE",
|
| 13 |
+
"106": "NOT_ALIVE",
|
| 14 |
"107": "NOT_ALIVE",
|
| 15 |
"108": "NOT_ALIVE",
|
| 16 |
+
"109": "NOT_ALIVE",
|
| 17 |
"11": "NOT_ALIVE",
|
| 18 |
"110": "NOT_ALIVE",
|
| 19 |
"111": "NON_PARTICIPATING",
|
|
|
|
| 22 |
"114": "NOT_ALIVE",
|
| 23 |
"115": "SUCCESS",
|
| 24 |
"116": "NON_PARTICIPATING",
|
| 25 |
+
"117": "NON_PARTICIPATING",
|
| 26 |
"118": "NOT_ALIVE",
|
| 27 |
"119": "NOT_ALIVE",
|
| 28 |
"12": "NOT_ALIVE",
|
| 29 |
"120": "NON_PARTICIPATING",
|
| 30 |
"121": "NOT_ALIVE",
|
| 31 |
"122": "NOT_ALIVE",
|
| 32 |
+
"123": "NOT_ALIVE",
|
| 33 |
"124": "SUCCESS",
|
| 34 |
"125": "SUCCESS",
|
| 35 |
"126": "NON_PARTICIPATING",
|
|
|
|
| 40 |
"130": "SUCCESS",
|
| 41 |
"131": "NOT_ALIVE",
|
| 42 |
"132": "NOT_ALIVE",
|
| 43 |
+
"133": "NON_PARTICIPATING",
|
| 44 |
"134": "SUCCESS",
|
| 45 |
+
"135": "SUCCESS",
|
| 46 |
"136": "NOT_ALIVE",
|
| 47 |
+
"137": "NON_PARTICIPATING",
|
| 48 |
"138": "NOT_ALIVE",
|
| 49 |
"139": "SUCCESS",
|
| 50 |
"14": "NON_PARTICIPATING",
|
|
|
|
| 53 |
"142": "SUCCESS",
|
| 54 |
"143": "NOT_ALIVE",
|
| 55 |
"144": "NOT_ALIVE",
|
| 56 |
+
"145": "NOT_ALIVE",
|
| 57 |
"146": "NOT_ALIVE",
|
| 58 |
"147": "NOT_ALIVE",
|
| 59 |
"148": "NOT_ALIVE",
|
| 60 |
+
"149": "SUCCESS",
|
| 61 |
"15": "NOT_ALIVE",
|
| 62 |
+
"150": "NON_PARTICIPATING",
|
| 63 |
"151": "NOT_ALIVE",
|
| 64 |
"152": "NON_PARTICIPATING",
|
| 65 |
+
"153": "SUCCESS",
|
| 66 |
"154": "NOT_ALIVE",
|
| 67 |
"155": "NOT_ALIVE",
|
| 68 |
"156": "NOT_ALIVE",
|
| 69 |
+
"157": "NON_PARTICIPATING",
|
| 70 |
"158": "SUCCESS",
|
| 71 |
"159": "SUCCESS",
|
| 72 |
+
"16": "NOT_ALIVE",
|
| 73 |
"160": "NON_PARTICIPATING",
|
| 74 |
+
"161": "NON_PARTICIPATING",
|
| 75 |
"162": "NON_PARTICIPATING",
|
| 76 |
"163": "NON_PARTICIPATING",
|
| 77 |
"164": "NOT_ALIVE",
|
| 78 |
+
"165": "NON_PARTICIPATING",
|
| 79 |
"166": "SUCCESS",
|
| 80 |
"167": "NOT_ALIVE",
|
| 81 |
"168": "SUCCESS",
|
| 82 |
"169": "SUCCESS",
|
| 83 |
"17": "NOT_ALIVE",
|
| 84 |
+
"170": "NON_PARTICIPATING",
|
| 85 |
"171": "NOT_ALIVE",
|
| 86 |
"172": "NOT_ALIVE",
|
| 87 |
+
"173": "NON_PARTICIPATING",
|
| 88 |
+
"174": "NON_PARTICIPATING",
|
| 89 |
+
"175": "SUCCESS",
|
| 90 |
"176": "NOT_ALIVE",
|
| 91 |
"177": "NON_PARTICIPATING",
|
| 92 |
"178": "NOT_ALIVE",
|
| 93 |
"179": "NOT_ALIVE",
|
| 94 |
+
"18": "NON_PARTICIPATING",
|
| 95 |
"180": "NON_PARTICIPATING",
|
| 96 |
"181": "NOT_ALIVE",
|
| 97 |
"182": "NOT_ALIVE",
|
| 98 |
"183": "NON_PARTICIPATING",
|
| 99 |
"184": "SUCCESS",
|
| 100 |
"185": "SUCCESS",
|
| 101 |
+
"186": "SUCCESS",
|
| 102 |
"187": "NOT_ALIVE",
|
| 103 |
"188": "NON_PARTICIPATING",
|
| 104 |
"189": "NON_PARTICIPATING",
|
| 105 |
"19": "SUCCESS",
|
| 106 |
+
"190": "NOT_ALIVE",
|
| 107 |
"191": "NOT_ALIVE",
|
| 108 |
"192": "NOT_ALIVE",
|
| 109 |
"193": "NOT_ALIVE",
|
|
|
|
| 111 |
"195": "NON_PARTICIPATING",
|
| 112 |
"196": "NON_PARTICIPATING",
|
| 113 |
"197": "NOT_ALIVE",
|
| 114 |
+
"198": "NON_PARTICIPATING",
|
| 115 |
+
"199": "NON_PARTICIPATING",
|
| 116 |
"2": "NOT_ALIVE",
|
| 117 |
"20": "NON_PARTICIPATING",
|
| 118 |
"200": "SUCCESS",
|
| 119 |
+
"201": "NOT_ALIVE",
|
| 120 |
"202": "NOT_ALIVE",
|
| 121 |
"203": "NOT_ALIVE",
|
| 122 |
"204": "NOT_ALIVE",
|
| 123 |
"205": "SUCCESS",
|
| 124 |
+
"206": "SUCCESS",
|
| 125 |
"207": "NOT_ALIVE",
|
| 126 |
"208": "NOT_ALIVE",
|
| 127 |
"209": "NOT_ALIVE",
|
| 128 |
"21": "SUCCESS",
|
| 129 |
"210": "NON_PARTICIPATING",
|
| 130 |
"211": "NOT_ALIVE",
|
| 131 |
+
"212": "NOT_ALIVE",
|
| 132 |
"213": "NON_PARTICIPATING",
|
| 133 |
"214": "NON_PARTICIPATING",
|
| 134 |
"215": "NOT_ALIVE",
|
| 135 |
+
"216": "NON_PARTICIPATING",
|
| 136 |
+
"217": "NON_PARTICIPATING",
|
| 137 |
"218": "NOT_ALIVE",
|
| 138 |
"219": "NON_PARTICIPATING",
|
| 139 |
"22": "NOT_ALIVE",
|
| 140 |
"220": "NON_PARTICIPATING",
|
| 141 |
+
"221": "NOT_ALIVE",
|
| 142 |
"222": "NOT_ALIVE",
|
| 143 |
+
"223": "NOT_ALIVE",
|
| 144 |
"224": "NOT_ALIVE",
|
| 145 |
"225": "SUCCESS",
|
| 146 |
"226": "SUCCESS",
|
|
|
|
| 152 |
"231": "NOT_ALIVE",
|
| 153 |
"232": "NOT_ALIVE",
|
| 154 |
"233": "NOT_ALIVE",
|
| 155 |
+
"234": "NOT_ALIVE",
|
| 156 |
"235": "NOT_ALIVE",
|
| 157 |
"236": "NON_PARTICIPATING",
|
| 158 |
"237": "SUCCESS",
|
| 159 |
+
"238": "NOT_ALIVE",
|
| 160 |
"239": "NOT_ALIVE",
|
| 161 |
"24": "SUCCESS",
|
| 162 |
"240": "NOT_ALIVE",
|
| 163 |
+
"241": "NOT_ALIVE",
|
| 164 |
"242": "NOT_ALIVE",
|
| 165 |
"243": "NOT_ALIVE",
|
| 166 |
+
"244": "NOT_ALIVE",
|
| 167 |
+
"245": "NON_PARTICIPATING",
|
| 168 |
"246": "NON_PARTICIPATING",
|
| 169 |
"247": "NON_PARTICIPATING",
|
| 170 |
"248": "SUCCESS",
|
| 171 |
"249": "NOT_ALIVE",
|
| 172 |
"25": "SUCCESS",
|
| 173 |
+
"250": "NON_PARTICIPATING",
|
| 174 |
"251": "NON_PARTICIPATING",
|
| 175 |
"252": "SUCCESS",
|
| 176 |
"253": "NON_PARTICIPATING",
|
|
|
|
| 181 |
"28": "NOT_ALIVE",
|
| 182 |
"29": "NOT_ALIVE",
|
| 183 |
"3": "SUCCESS",
|
| 184 |
+
"30": "NON_PARTICIPATING",
|
| 185 |
"31": "NON_PARTICIPATING",
|
| 186 |
"32": "NOT_ALIVE",
|
| 187 |
"33": "NOT_ALIVE",
|
| 188 |
"34": "NON_PARTICIPATING",
|
| 189 |
+
"35": "SUCCESS",
|
| 190 |
+
"36": "NOT_ALIVE",
|
| 191 |
"37": "NOT_ALIVE",
|
| 192 |
"38": "NON_PARTICIPATING",
|
| 193 |
"39": "SUCCESS",
|
| 194 |
"4": "NOT_ALIVE",
|
| 195 |
"40": "NOT_ALIVE",
|
| 196 |
+
"41": "NOT_ALIVE",
|
| 197 |
"42": "NOT_ALIVE",
|
| 198 |
"43": "NON_PARTICIPATING",
|
| 199 |
"44": "NON_PARTICIPATING",
|
|
|
|
| 207 |
"51": "NON_PARTICIPATING",
|
| 208 |
"52": "SUCCESS",
|
| 209 |
"53": "SUCCESS",
|
| 210 |
+
"54": "NON_PARTICIPATING",
|
| 211 |
"55": "NOT_ALIVE",
|
| 212 |
"56": "NOT_ALIVE",
|
| 213 |
+
"57": "SUCCESS",
|
| 214 |
"58": "NON_PARTICIPATING",
|
| 215 |
"59": "NON_PARTICIPATING",
|
| 216 |
"6": "NOT_ALIVE",
|
| 217 |
"60": "NON_PARTICIPATING",
|
| 218 |
"61": "SUCCESS",
|
| 219 |
"62": "SUCCESS",
|
| 220 |
+
"63": "NOT_ALIVE",
|
| 221 |
"64": "SUCCESS",
|
| 222 |
"65": "NOT_ALIVE",
|
| 223 |
"66": "NOT_ALIVE",
|
|
|
|
| 228 |
"70": "NOT_ALIVE",
|
| 229 |
"71": "NOT_ALIVE",
|
| 230 |
"72": "NOT_ALIVE",
|
| 231 |
+
"73": "NOT_ALIVE",
|
| 232 |
"74": "NOT_ALIVE",
|
| 233 |
"75": "NON_PARTICIPATING",
|
| 234 |
"76": "SUCCESS",
|
|
|
|
| 241 |
"82": "NOT_ALIVE",
|
| 242 |
"83": "NOT_ALIVE",
|
| 243 |
"84": "NON_PARTICIPATING",
|
| 244 |
+
"85": "NOT_ALIVE",
|
| 245 |
"86": "NON_PARTICIPATING",
|
| 246 |
"87": "SUCCESS",
|
| 247 |
"88": "NOT_ALIVE",
|
|
|
|
| 253 |
"93": "SUCCESS",
|
| 254 |
"94": "NON_PARTICIPATING",
|
| 255 |
"95": "NOT_ALIVE",
|
| 256 |
+
"96": "NON_PARTICIPATING",
|
| 257 |
"97": "NOT_ALIVE",
|
| 258 |
+
"98": "SUCCESS",
|
| 259 |
"99": "SUCCESS"
|
| 260 |
},
|
| 261 |
"architectures": [
|
|
|
|
| 264 |
"attention_bias": false,
|
| 265 |
"attention_dropout": 0.0,
|
| 266 |
"block_list": [
|
| 267 |
+
6581561,
|
| 268 |
+
6581590
|
| 269 |
],
|
| 270 |
"bos_token_id": 1,
|
| 271 |
"dtype": "float32",
|
|
|
|
| 274 |
"hidden_act": "silu",
|
| 275 |
"hidden_size": 2048,
|
| 276 |
"initializer_range": 0.02,
|
| 277 |
+
"inner_step": 9,
|
| 278 |
"intermediate_size": 5632,
|
| 279 |
+
"last_allreduce_block": 6581540,
|
| 280 |
"max_position_embeddings": 2048,
|
| 281 |
"mlp_bias": false,
|
| 282 |
"model_type": "llama",
|
gradients.pt
CHANGED
|
@@ -1,3 +1,3 @@
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
-
oid sha256:
|
| 3 |
size 88390091
|
|
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:583251fdf02383d6b109c253c5c467ade9034f82efc459497a5c668eb28c283a
|
| 3 |
size 88390091
|
inner_optimizer.pt
CHANGED
|
@@ -1,3 +1,3 @@
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
-
oid sha256:
|
| 3 |
size 4924928111
|
|
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:176f6b4b4d308fae5b6e181bbd65b8657daa149a6473cf225afa231a086c4f1b
|
| 3 |
size 4924928111
|
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 sha256:9d8531d9445b4149e79d78e3e420dfb7a50d1e77b3d1be77bc34aab7aa3eb311
|
| 3 |
size 4400216536
|