Run 7. Outer Step 128. Inner Step 4.
Browse files- config.json +47 -47
- gradients.pt +1 -1
- inner_optimizer.pt +1 -1
- model.safetensors +1 -1
config.json
CHANGED
|
@@ -9,26 +9,26 @@
|
|
| 9 |
"102": "NOT_ALIVE",
|
| 10 |
"103": "NOT_ALIVE",
|
| 11 |
"104": "NOT_ALIVE",
|
| 12 |
-
"105": "
|
| 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": "
|
| 20 |
"112": "NON_PARTICIPATING",
|
| 21 |
"113": "SUCCESS",
|
| 22 |
"114": "NOT_ALIVE",
|
| 23 |
"115": "NOT_ALIVE",
|
| 24 |
-
"116": "
|
| 25 |
-
"117": "
|
| 26 |
"118": "NOT_ALIVE",
|
| 27 |
"119": "NON_PARTICIPATING",
|
| 28 |
"12": "NOT_ALIVE",
|
| 29 |
"120": "NOT_ALIVE",
|
| 30 |
"121": "NOT_ALIVE",
|
| 31 |
-
"122": "
|
| 32 |
"123": "NON_PARTICIPATING",
|
| 33 |
"124": "NON_PARTICIPATING",
|
| 34 |
"125": "NOT_ALIVE",
|
|
@@ -37,11 +37,11 @@
|
|
| 37 |
"128": "NOT_ALIVE",
|
| 38 |
"129": "NOT_ALIVE",
|
| 39 |
"13": "NOT_ALIVE",
|
| 40 |
-
"130": "
|
| 41 |
"131": "NOT_ALIVE",
|
| 42 |
"132": "NOT_ALIVE",
|
| 43 |
"133": "SUCCESS",
|
| 44 |
-
"134": "
|
| 45 |
"135": "NOT_ALIVE",
|
| 46 |
"136": "NON_PARTICIPATING",
|
| 47 |
"137": "NON_PARTICIPATING",
|
|
@@ -50,7 +50,7 @@
|
|
| 50 |
"14": "SUCCESS",
|
| 51 |
"140": "NON_PARTICIPATING",
|
| 52 |
"141": "NOT_ALIVE",
|
| 53 |
-
"142": "
|
| 54 |
"143": "SUCCESS",
|
| 55 |
"144": "SUCCESS",
|
| 56 |
"145": "NOT_ALIVE",
|
|
@@ -61,52 +61,52 @@
|
|
| 61 |
"15": "SUCCESS",
|
| 62 |
"150": "NOT_ALIVE",
|
| 63 |
"151": "NOT_ALIVE",
|
| 64 |
-
"152": "
|
| 65 |
-
"153": "
|
| 66 |
-
"154": "
|
| 67 |
"155": "NON_PARTICIPATING",
|
| 68 |
"156": "NOT_ALIVE",
|
| 69 |
"157": "NON_PARTICIPATING",
|
| 70 |
"158": "NON_PARTICIPATING",
|
| 71 |
"159": "NON_PARTICIPATING",
|
| 72 |
"16": "SUCCESS",
|
| 73 |
-
"160": "
|
| 74 |
"161": "NOT_ALIVE",
|
| 75 |
"162": "NOT_ALIVE",
|
| 76 |
"163": "SUCCESS",
|
| 77 |
"164": "NON_PARTICIPATING",
|
| 78 |
-
"165": "
|
| 79 |
"166": "NOT_ALIVE",
|
| 80 |
-
"167": "
|
| 81 |
"168": "NON_PARTICIPATING",
|
| 82 |
"169": "NON_PARTICIPATING",
|
| 83 |
"17": "NOT_ALIVE",
|
| 84 |
"170": "NON_PARTICIPATING",
|
| 85 |
"171": "NOT_ALIVE",
|
| 86 |
-
"172": "
|
| 87 |
"173": "SUCCESS",
|
| 88 |
"174": "NOT_ALIVE",
|
| 89 |
"175": "NON_PARTICIPATING",
|
| 90 |
"176": "NOT_ALIVE",
|
| 91 |
"177": "NOT_ALIVE",
|
| 92 |
"178": "NOT_ALIVE",
|
| 93 |
-
"179": "
|
| 94 |
-
"18": "
|
| 95 |
"180": "NON_PARTICIPATING",
|
| 96 |
"181": "SUCCESS",
|
| 97 |
-
"182": "
|
| 98 |
"183": "NON_PARTICIPATING",
|
| 99 |
"184": "NOT_ALIVE",
|
| 100 |
"185": "NON_PARTICIPATING",
|
| 101 |
-
"186": "
|
| 102 |
"187": "SUCCESS",
|
| 103 |
"188": "NON_PARTICIPATING",
|
| 104 |
"189": "SUCCESS",
|
| 105 |
"19": "NOT_ALIVE",
|
| 106 |
"190": "NOT_ALIVE",
|
| 107 |
"191": "SUCCESS",
|
| 108 |
-
"192": "
|
| 109 |
-
"193": "
|
| 110 |
"194": "SUCCESS",
|
| 111 |
"195": "NOT_ALIVE",
|
| 112 |
"196": "NOT_ALIVE",
|
|
@@ -118,21 +118,21 @@
|
|
| 118 |
"200": "NOT_ALIVE",
|
| 119 |
"201": "NOT_ALIVE",
|
| 120 |
"202": "NON_PARTICIPATING",
|
| 121 |
-
"203": "
|
| 122 |
"204": "NON_PARTICIPATING",
|
| 123 |
"205": "NOT_ALIVE",
|
| 124 |
"206": "NON_PARTICIPATING",
|
| 125 |
"207": "NOT_ALIVE",
|
| 126 |
"208": "NOT_ALIVE",
|
| 127 |
"209": "NOT_ALIVE",
|
| 128 |
-
"21": "
|
| 129 |
"210": "NON_PARTICIPATING",
|
| 130 |
"211": "SUCCESS",
|
| 131 |
"212": "NON_PARTICIPATING",
|
| 132 |
-
"213": "
|
| 133 |
-
"214": "
|
| 134 |
"215": "NOT_ALIVE",
|
| 135 |
-
"216": "
|
| 136 |
"217": "NON_PARTICIPATING",
|
| 137 |
"218": "NON_PARTICIPATING",
|
| 138 |
"219": "SUCCESS",
|
|
@@ -153,7 +153,7 @@
|
|
| 153 |
"232": "NOT_ALIVE",
|
| 154 |
"233": "NOT_ALIVE",
|
| 155 |
"234": "SUCCESS",
|
| 156 |
-
"235": "
|
| 157 |
"236": "NON_PARTICIPATING",
|
| 158 |
"237": "NON_PARTICIPATING",
|
| 159 |
"238": "NOT_ALIVE",
|
|
@@ -161,11 +161,11 @@
|
|
| 161 |
"24": "NOT_ALIVE",
|
| 162 |
"240": "NOT_ALIVE",
|
| 163 |
"241": "NON_PARTICIPATING",
|
| 164 |
-
"242": "
|
| 165 |
"243": "NOT_ALIVE",
|
| 166 |
"244": "NON_PARTICIPATING",
|
| 167 |
"245": "NOT_ALIVE",
|
| 168 |
-
"246": "
|
| 169 |
"247": "NON_PARTICIPATING",
|
| 170 |
"248": "NOT_ALIVE",
|
| 171 |
"249": "NOT_ALIVE",
|
|
@@ -173,14 +173,14 @@
|
|
| 173 |
"250": "NON_PARTICIPATING",
|
| 174 |
"251": "NOT_ALIVE",
|
| 175 |
"252": "NOT_ALIVE",
|
| 176 |
-
"253": "
|
| 177 |
"254": "NOT_ALIVE",
|
| 178 |
"255": "NOT_ALIVE",
|
| 179 |
-
"26": "
|
| 180 |
"27": "SUCCESS",
|
| 181 |
"28": "NON_PARTICIPATING",
|
| 182 |
"29": "NOT_ALIVE",
|
| 183 |
-
"3": "
|
| 184 |
"30": "NOT_ALIVE",
|
| 185 |
"31": "NON_PARTICIPATING",
|
| 186 |
"32": "NOT_ALIVE",
|
|
@@ -195,7 +195,7 @@
|
|
| 195 |
"40": "NON_PARTICIPATING",
|
| 196 |
"41": "NON_PARTICIPATING",
|
| 197 |
"42": "NON_PARTICIPATING",
|
| 198 |
-
"43": "
|
| 199 |
"44": "NOT_ALIVE",
|
| 200 |
"45": "NOT_ALIVE",
|
| 201 |
"46": "NON_PARTICIPATING",
|
|
@@ -205,17 +205,17 @@
|
|
| 205 |
"5": "NOT_ALIVE",
|
| 206 |
"50": "NOT_ALIVE",
|
| 207 |
"51": "NOT_ALIVE",
|
| 208 |
-
"52": "
|
| 209 |
"53": "NON_PARTICIPATING",
|
| 210 |
"54": "NOT_ALIVE",
|
| 211 |
"55": "NON_PARTICIPATING",
|
| 212 |
"56": "NOT_ALIVE",
|
| 213 |
"57": "NOT_ALIVE",
|
| 214 |
"58": "NOT_ALIVE",
|
| 215 |
-
"59": "
|
| 216 |
"6": "NOT_ALIVE",
|
| 217 |
"60": "NON_PARTICIPATING",
|
| 218 |
-
"61": "
|
| 219 |
"62": "NOT_ALIVE",
|
| 220 |
"63": "NOT_ALIVE",
|
| 221 |
"64": "SUCCESS",
|
|
@@ -223,7 +223,7 @@
|
|
| 223 |
"66": "NOT_ALIVE",
|
| 224 |
"67": "NOT_ALIVE",
|
| 225 |
"68": "NOT_ALIVE",
|
| 226 |
-
"69": "
|
| 227 |
"7": "NOT_ALIVE",
|
| 228 |
"70": "NOT_ALIVE",
|
| 229 |
"71": "NOT_ALIVE",
|
|
@@ -232,15 +232,15 @@
|
|
| 232 |
"74": "NOT_ALIVE",
|
| 233 |
"75": "NOT_ALIVE",
|
| 234 |
"76": "NOT_ALIVE",
|
| 235 |
-
"77": "
|
| 236 |
"78": "NON_PARTICIPATING",
|
| 237 |
"79": "NOT_ALIVE",
|
| 238 |
"8": "NOT_ALIVE",
|
| 239 |
-
"80": "
|
| 240 |
"81": "NOT_ALIVE",
|
| 241 |
"82": "NON_PARTICIPATING",
|
| 242 |
"83": "NOT_ALIVE",
|
| 243 |
-
"84": "
|
| 244 |
"85": "NOT_ALIVE",
|
| 245 |
"86": "NON_PARTICIPATING",
|
| 246 |
"87": "NOT_ALIVE",
|
|
@@ -249,13 +249,13 @@
|
|
| 249 |
"9": "NOT_ALIVE",
|
| 250 |
"90": "NOT_ALIVE",
|
| 251 |
"91": "NOT_ALIVE",
|
| 252 |
-
"92": "
|
| 253 |
"93": "NOT_ALIVE",
|
| 254 |
-
"94": "
|
| 255 |
"95": "NOT_ALIVE",
|
| 256 |
"96": "NOT_ALIVE",
|
| 257 |
"97": "NOT_ALIVE",
|
| 258 |
-
"98": "
|
| 259 |
"99": "NOT_ALIVE"
|
| 260 |
},
|
| 261 |
"architectures": [
|
|
@@ -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": "NOT_ALIVE",
|
| 11 |
"104": "NOT_ALIVE",
|
| 12 |
+
"105": "NON_PARTICIPATING",
|
| 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",
|
| 20 |
"112": "NON_PARTICIPATING",
|
| 21 |
"113": "SUCCESS",
|
| 22 |
"114": "NOT_ALIVE",
|
| 23 |
"115": "NOT_ALIVE",
|
| 24 |
+
"116": "NON_PARTICIPATING",
|
| 25 |
+
"117": "NON_PARTICIPATING",
|
| 26 |
"118": "NOT_ALIVE",
|
| 27 |
"119": "NON_PARTICIPATING",
|
| 28 |
"12": "NOT_ALIVE",
|
| 29 |
"120": "NOT_ALIVE",
|
| 30 |
"121": "NOT_ALIVE",
|
| 31 |
+
"122": "NON_PARTICIPATING",
|
| 32 |
"123": "NON_PARTICIPATING",
|
| 33 |
"124": "NON_PARTICIPATING",
|
| 34 |
"125": "NOT_ALIVE",
|
|
|
|
| 37 |
"128": "NOT_ALIVE",
|
| 38 |
"129": "NOT_ALIVE",
|
| 39 |
"13": "NOT_ALIVE",
|
| 40 |
+
"130": "SUCCESS",
|
| 41 |
"131": "NOT_ALIVE",
|
| 42 |
"132": "NOT_ALIVE",
|
| 43 |
"133": "SUCCESS",
|
| 44 |
+
"134": "SUCCESS",
|
| 45 |
"135": "NOT_ALIVE",
|
| 46 |
"136": "NON_PARTICIPATING",
|
| 47 |
"137": "NON_PARTICIPATING",
|
|
|
|
| 50 |
"14": "SUCCESS",
|
| 51 |
"140": "NON_PARTICIPATING",
|
| 52 |
"141": "NOT_ALIVE",
|
| 53 |
+
"142": "NON_PARTICIPATING",
|
| 54 |
"143": "SUCCESS",
|
| 55 |
"144": "SUCCESS",
|
| 56 |
"145": "NOT_ALIVE",
|
|
|
|
| 61 |
"15": "SUCCESS",
|
| 62 |
"150": "NOT_ALIVE",
|
| 63 |
"151": "NOT_ALIVE",
|
| 64 |
+
"152": "NON_PARTICIPATING",
|
| 65 |
+
"153": "NON_PARTICIPATING",
|
| 66 |
+
"154": "NON_PARTICIPATING",
|
| 67 |
"155": "NON_PARTICIPATING",
|
| 68 |
"156": "NOT_ALIVE",
|
| 69 |
"157": "NON_PARTICIPATING",
|
| 70 |
"158": "NON_PARTICIPATING",
|
| 71 |
"159": "NON_PARTICIPATING",
|
| 72 |
"16": "SUCCESS",
|
| 73 |
+
"160": "NOT_ALIVE",
|
| 74 |
"161": "NOT_ALIVE",
|
| 75 |
"162": "NOT_ALIVE",
|
| 76 |
"163": "SUCCESS",
|
| 77 |
"164": "NON_PARTICIPATING",
|
| 78 |
+
"165": "NOT_ALIVE",
|
| 79 |
"166": "NOT_ALIVE",
|
| 80 |
+
"167": "NON_PARTICIPATING",
|
| 81 |
"168": "NON_PARTICIPATING",
|
| 82 |
"169": "NON_PARTICIPATING",
|
| 83 |
"17": "NOT_ALIVE",
|
| 84 |
"170": "NON_PARTICIPATING",
|
| 85 |
"171": "NOT_ALIVE",
|
| 86 |
+
"172": "NOT_ALIVE",
|
| 87 |
"173": "SUCCESS",
|
| 88 |
"174": "NOT_ALIVE",
|
| 89 |
"175": "NON_PARTICIPATING",
|
| 90 |
"176": "NOT_ALIVE",
|
| 91 |
"177": "NOT_ALIVE",
|
| 92 |
"178": "NOT_ALIVE",
|
| 93 |
+
"179": "SUCCESS",
|
| 94 |
+
"18": "SUCCESS",
|
| 95 |
"180": "NON_PARTICIPATING",
|
| 96 |
"181": "SUCCESS",
|
| 97 |
+
"182": "NON_PARTICIPATING",
|
| 98 |
"183": "NON_PARTICIPATING",
|
| 99 |
"184": "NOT_ALIVE",
|
| 100 |
"185": "NON_PARTICIPATING",
|
| 101 |
+
"186": "NOT_ALIVE",
|
| 102 |
"187": "SUCCESS",
|
| 103 |
"188": "NON_PARTICIPATING",
|
| 104 |
"189": "SUCCESS",
|
| 105 |
"19": "NOT_ALIVE",
|
| 106 |
"190": "NOT_ALIVE",
|
| 107 |
"191": "SUCCESS",
|
| 108 |
+
"192": "NON_PARTICIPATING",
|
| 109 |
+
"193": "NON_PARTICIPATING",
|
| 110 |
"194": "SUCCESS",
|
| 111 |
"195": "NOT_ALIVE",
|
| 112 |
"196": "NOT_ALIVE",
|
|
|
|
| 118 |
"200": "NOT_ALIVE",
|
| 119 |
"201": "NOT_ALIVE",
|
| 120 |
"202": "NON_PARTICIPATING",
|
| 121 |
+
"203": "NOT_ALIVE",
|
| 122 |
"204": "NON_PARTICIPATING",
|
| 123 |
"205": "NOT_ALIVE",
|
| 124 |
"206": "NON_PARTICIPATING",
|
| 125 |
"207": "NOT_ALIVE",
|
| 126 |
"208": "NOT_ALIVE",
|
| 127 |
"209": "NOT_ALIVE",
|
| 128 |
+
"21": "NOT_ALIVE",
|
| 129 |
"210": "NON_PARTICIPATING",
|
| 130 |
"211": "SUCCESS",
|
| 131 |
"212": "NON_PARTICIPATING",
|
| 132 |
+
"213": "NOT_ALIVE",
|
| 133 |
+
"214": "SUCCESS",
|
| 134 |
"215": "NOT_ALIVE",
|
| 135 |
+
"216": "NON_PARTICIPATING",
|
| 136 |
"217": "NON_PARTICIPATING",
|
| 137 |
"218": "NON_PARTICIPATING",
|
| 138 |
"219": "SUCCESS",
|
|
|
|
| 153 |
"232": "NOT_ALIVE",
|
| 154 |
"233": "NOT_ALIVE",
|
| 155 |
"234": "SUCCESS",
|
| 156 |
+
"235": "NON_PARTICIPATING",
|
| 157 |
"236": "NON_PARTICIPATING",
|
| 158 |
"237": "NON_PARTICIPATING",
|
| 159 |
"238": "NOT_ALIVE",
|
|
|
|
| 161 |
"24": "NOT_ALIVE",
|
| 162 |
"240": "NOT_ALIVE",
|
| 163 |
"241": "NON_PARTICIPATING",
|
| 164 |
+
"242": "NON_PARTICIPATING",
|
| 165 |
"243": "NOT_ALIVE",
|
| 166 |
"244": "NON_PARTICIPATING",
|
| 167 |
"245": "NOT_ALIVE",
|
| 168 |
+
"246": "SUCCESS",
|
| 169 |
"247": "NON_PARTICIPATING",
|
| 170 |
"248": "NOT_ALIVE",
|
| 171 |
"249": "NOT_ALIVE",
|
|
|
|
| 173 |
"250": "NON_PARTICIPATING",
|
| 174 |
"251": "NOT_ALIVE",
|
| 175 |
"252": "NOT_ALIVE",
|
| 176 |
+
"253": "SUCCESS",
|
| 177 |
"254": "NOT_ALIVE",
|
| 178 |
"255": "NOT_ALIVE",
|
| 179 |
+
"26": "NOT_ALIVE",
|
| 180 |
"27": "SUCCESS",
|
| 181 |
"28": "NON_PARTICIPATING",
|
| 182 |
"29": "NOT_ALIVE",
|
| 183 |
+
"3": "NOT_ALIVE",
|
| 184 |
"30": "NOT_ALIVE",
|
| 185 |
"31": "NON_PARTICIPATING",
|
| 186 |
"32": "NOT_ALIVE",
|
|
|
|
| 195 |
"40": "NON_PARTICIPATING",
|
| 196 |
"41": "NON_PARTICIPATING",
|
| 197 |
"42": "NON_PARTICIPATING",
|
| 198 |
+
"43": "NON_PARTICIPATING",
|
| 199 |
"44": "NOT_ALIVE",
|
| 200 |
"45": "NOT_ALIVE",
|
| 201 |
"46": "NON_PARTICIPATING",
|
|
|
|
| 205 |
"5": "NOT_ALIVE",
|
| 206 |
"50": "NOT_ALIVE",
|
| 207 |
"51": "NOT_ALIVE",
|
| 208 |
+
"52": "NON_PARTICIPATING",
|
| 209 |
"53": "NON_PARTICIPATING",
|
| 210 |
"54": "NOT_ALIVE",
|
| 211 |
"55": "NON_PARTICIPATING",
|
| 212 |
"56": "NOT_ALIVE",
|
| 213 |
"57": "NOT_ALIVE",
|
| 214 |
"58": "NOT_ALIVE",
|
| 215 |
+
"59": "SUCCESS",
|
| 216 |
"6": "NOT_ALIVE",
|
| 217 |
"60": "NON_PARTICIPATING",
|
| 218 |
+
"61": "NOT_ALIVE",
|
| 219 |
"62": "NOT_ALIVE",
|
| 220 |
"63": "NOT_ALIVE",
|
| 221 |
"64": "SUCCESS",
|
|
|
|
| 223 |
"66": "NOT_ALIVE",
|
| 224 |
"67": "NOT_ALIVE",
|
| 225 |
"68": "NOT_ALIVE",
|
| 226 |
+
"69": "NON_PARTICIPATING",
|
| 227 |
"7": "NOT_ALIVE",
|
| 228 |
"70": "NOT_ALIVE",
|
| 229 |
"71": "NOT_ALIVE",
|
|
|
|
| 232 |
"74": "NOT_ALIVE",
|
| 233 |
"75": "NOT_ALIVE",
|
| 234 |
"76": "NOT_ALIVE",
|
| 235 |
+
"77": "NOT_ALIVE",
|
| 236 |
"78": "NON_PARTICIPATING",
|
| 237 |
"79": "NOT_ALIVE",
|
| 238 |
"8": "NOT_ALIVE",
|
| 239 |
+
"80": "SUCCESS",
|
| 240 |
"81": "NOT_ALIVE",
|
| 241 |
"82": "NON_PARTICIPATING",
|
| 242 |
"83": "NOT_ALIVE",
|
| 243 |
+
"84": "NON_PARTICIPATING",
|
| 244 |
"85": "NOT_ALIVE",
|
| 245 |
"86": "NON_PARTICIPATING",
|
| 246 |
"87": "NOT_ALIVE",
|
|
|
|
| 249 |
"9": "NOT_ALIVE",
|
| 250 |
"90": "NOT_ALIVE",
|
| 251 |
"91": "NOT_ALIVE",
|
| 252 |
+
"92": "SUCCESS",
|
| 253 |
"93": "NOT_ALIVE",
|
| 254 |
+
"94": "NON_PARTICIPATING",
|
| 255 |
"95": "NOT_ALIVE",
|
| 256 |
"96": "NOT_ALIVE",
|
| 257 |
"97": "NOT_ALIVE",
|
| 258 |
+
"98": "NOT_ALIVE",
|
| 259 |
"99": "NOT_ALIVE"
|
| 260 |
},
|
| 261 |
"architectures": [
|
|
|
|
| 264 |
"attention_bias": false,
|
| 265 |
"attention_dropout": 0.0,
|
| 266 |
"block_list": [
|
| 267 |
+
6313450,
|
| 268 |
+
6313468
|
| 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": 4,
|
| 276 |
"intermediate_size": 5632,
|
| 277 |
+
"last_allreduce_block": 6313415,
|
| 278 |
"max_position_embeddings": 2048,
|
| 279 |
"mlp_bias": false,
|
| 280 |
"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 sha256:93aba177dd54a1d78e02c5ff5b4410989766f33b9f81b272d33ac1eb49e0fdfc
|
| 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 8800564561
|
|
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:ffeefd60b4b92011abb3d84bcfc38101b00634fe5d8d969c1fb3bccf631378ce
|
| 3 |
size 8800564561
|
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 sha256:58dbc9ff42d98fdf484dfe69d75757482ef55b51c20b3f2cac0634e8f689071b
|
| 3 |
size 4400216536
|