Run 8. Outer Step 49. Inner Step 5.
Browse files- config.json +54 -56
- gradients.pt +1 -1
- inner_optimizer.pt +1 -1
- model.safetensors +1 -1
config.json
CHANGED
|
@@ -4,7 +4,7 @@
|
|
| 4 |
"0": "NOT_ALIVE",
|
| 5 |
"1": "NOT_ALIVE",
|
| 6 |
"10": "NOT_ALIVE",
|
| 7 |
-
"100": "
|
| 8 |
"101": "NOT_ALIVE",
|
| 9 |
"102": "NON_PARTICIPATING",
|
| 10 |
"103": "NOT_ALIVE",
|
|
@@ -17,10 +17,10 @@
|
|
| 17 |
"11": "NOT_ALIVE",
|
| 18 |
"110": "NON_PARTICIPATING",
|
| 19 |
"111": "NOT_ALIVE",
|
| 20 |
-
"112": "
|
| 21 |
"113": "NOT_ALIVE",
|
| 22 |
"114": "NOT_ALIVE",
|
| 23 |
-
"115": "
|
| 24 |
"116": "NOT_ALIVE",
|
| 25 |
"117": "NOT_ALIVE",
|
| 26 |
"118": "NOT_ALIVE",
|
|
@@ -32,26 +32,26 @@
|
|
| 32 |
"123": "SUCCESS",
|
| 33 |
"124": "NOT_ALIVE",
|
| 34 |
"125": "NOT_ALIVE",
|
| 35 |
-
"126": "
|
| 36 |
-
"127": "
|
| 37 |
"128": "SUCCESS",
|
| 38 |
-
"129": "
|
| 39 |
"13": "NOT_ALIVE",
|
| 40 |
-
"130": "
|
| 41 |
"131": "NOT_ALIVE",
|
| 42 |
"132": "SUCCESS",
|
| 43 |
"133": "SUCCESS",
|
| 44 |
-
"134": "
|
| 45 |
"135": "SUCCESS",
|
| 46 |
"136": "SUCCESS",
|
| 47 |
-
"137": "
|
| 48 |
"138": "NON_PARTICIPATING",
|
| 49 |
"139": "NOT_ALIVE",
|
| 50 |
"14": "NOT_ALIVE",
|
| 51 |
"140": "NON_PARTICIPATING",
|
| 52 |
"141": "NOT_ALIVE",
|
| 53 |
"142": "NON_PARTICIPATING",
|
| 54 |
-
"143": "
|
| 55 |
"144": "NON_PARTICIPATING",
|
| 56 |
"145": "SUCCESS",
|
| 57 |
"146": "NON_PARTICIPATING",
|
|
@@ -59,12 +59,12 @@
|
|
| 59 |
"148": "NON_PARTICIPATING",
|
| 60 |
"149": "SUCCESS",
|
| 61 |
"15": "SUCCESS",
|
| 62 |
-
"150": "
|
| 63 |
"151": "NON_PARTICIPATING",
|
| 64 |
"152": "FAIL",
|
| 65 |
"153": "SUCCESS",
|
| 66 |
"154": "NOT_ALIVE",
|
| 67 |
-
"155": "
|
| 68 |
"156": "NOT_ALIVE",
|
| 69 |
"157": "NON_PARTICIPATING",
|
| 70 |
"158": "NON_PARTICIPATING",
|
|
@@ -72,8 +72,8 @@
|
|
| 72 |
"16": "NOT_ALIVE",
|
| 73 |
"160": "NON_PARTICIPATING",
|
| 74 |
"161": "NON_PARTICIPATING",
|
| 75 |
-
"162": "
|
| 76 |
-
"163": "
|
| 77 |
"164": "FAIL",
|
| 78 |
"165": "SUCCESS",
|
| 79 |
"166": "NOT_ALIVE",
|
|
@@ -86,7 +86,7 @@
|
|
| 86 |
"172": "NON_PARTICIPATING",
|
| 87 |
"173": "NOT_ALIVE",
|
| 88 |
"174": "SUCCESS",
|
| 89 |
-
"175": "
|
| 90 |
"176": "NOT_ALIVE",
|
| 91 |
"177": "SUCCESS",
|
| 92 |
"178": "NON_PARTICIPATING",
|
|
@@ -95,33 +95,33 @@
|
|
| 95 |
"180": "NON_PARTICIPATING",
|
| 96 |
"181": "FAIL",
|
| 97 |
"182": "NOT_ALIVE",
|
| 98 |
-
"183": "
|
| 99 |
"184": "NOT_ALIVE",
|
| 100 |
-
"185": "
|
| 101 |
"186": "FAIL",
|
| 102 |
"187": "NOT_ALIVE",
|
| 103 |
"188": "FAIL",
|
| 104 |
"189": "SUCCESS",
|
| 105 |
"19": "NOT_ALIVE",
|
| 106 |
-
"190": "
|
| 107 |
-
"191": "
|
| 108 |
-
"192": "
|
| 109 |
"193": "NOT_ALIVE",
|
| 110 |
"194": "NON_PARTICIPATING",
|
| 111 |
"195": "NON_PARTICIPATING",
|
| 112 |
"196": "SUCCESS",
|
| 113 |
"197": "NOT_ALIVE",
|
| 114 |
"198": "NOT_ALIVE",
|
| 115 |
-
"199": "
|
| 116 |
"2": "NOT_ALIVE",
|
| 117 |
"20": "NOT_ALIVE",
|
| 118 |
-
"200": "
|
| 119 |
"201": "NOT_ALIVE",
|
| 120 |
"202": "FAIL",
|
| 121 |
"203": "NOT_ALIVE",
|
| 122 |
"204": "NON_PARTICIPATING",
|
| 123 |
"205": "NOT_ALIVE",
|
| 124 |
-
"206": "
|
| 125 |
"207": "NOT_ALIVE",
|
| 126 |
"208": "NOT_ALIVE",
|
| 127 |
"209": "NOT_ALIVE",
|
|
@@ -137,11 +137,11 @@
|
|
| 137 |
"218": "FAIL",
|
| 138 |
"219": "FAIL",
|
| 139 |
"22": "NOT_ALIVE",
|
| 140 |
-
"220": "
|
| 141 |
"221": "SUCCESS",
|
| 142 |
"222": "NOT_ALIVE",
|
| 143 |
"223": "NOT_ALIVE",
|
| 144 |
-
"224": "
|
| 145 |
"225": "SUCCESS",
|
| 146 |
"226": "NOT_ALIVE",
|
| 147 |
"227": "NON_PARTICIPATING",
|
|
@@ -149,14 +149,14 @@
|
|
| 149 |
"229": "NON_PARTICIPATING",
|
| 150 |
"23": "NOT_ALIVE",
|
| 151 |
"230": "NOT_ALIVE",
|
| 152 |
-
"231": "
|
| 153 |
"232": "NOT_ALIVE",
|
| 154 |
"233": "NON_PARTICIPATING",
|
| 155 |
"234": "SUCCESS",
|
| 156 |
"235": "NOT_ALIVE",
|
| 157 |
"236": "SUCCESS",
|
| 158 |
"237": "NON_PARTICIPATING",
|
| 159 |
-
"238": "
|
| 160 |
"239": "NON_PARTICIPATING",
|
| 161 |
"24": "SUCCESS",
|
| 162 |
"240": "NON_PARTICIPATING",
|
|
@@ -164,7 +164,7 @@
|
|
| 164 |
"242": "NON_PARTICIPATING",
|
| 165 |
"243": "NON_PARTICIPATING",
|
| 166 |
"244": "NON_PARTICIPATING",
|
| 167 |
-
"245": "
|
| 168 |
"246": "NON_PARTICIPATING",
|
| 169 |
"247": "NON_PARTICIPATING",
|
| 170 |
"248": "NON_PARTICIPATING",
|
|
@@ -172,46 +172,46 @@
|
|
| 172 |
"25": "SUCCESS",
|
| 173 |
"250": "NON_PARTICIPATING",
|
| 174 |
"251": "NON_PARTICIPATING",
|
| 175 |
-
"252": "
|
| 176 |
"253": "SUCCESS",
|
| 177 |
"254": "NOT_ALIVE",
|
| 178 |
"255": "NON_PARTICIPATING",
|
| 179 |
-
"26": "
|
| 180 |
"27": "SUCCESS",
|
| 181 |
"28": "NON_PARTICIPATING",
|
| 182 |
"29": "SUCCESS",
|
| 183 |
"3": "FAIL",
|
| 184 |
"30": "SUCCESS",
|
| 185 |
-
"31": "
|
| 186 |
"32": "NON_PARTICIPATING",
|
| 187 |
"33": "NOT_ALIVE",
|
| 188 |
"34": "NOT_ALIVE",
|
| 189 |
-
"35": "
|
| 190 |
"36": "FAIL",
|
| 191 |
"37": "NOT_ALIVE",
|
| 192 |
"38": "SUCCESS",
|
| 193 |
-
"39": "
|
| 194 |
"4": "NOT_ALIVE",
|
| 195 |
"40": "NON_PARTICIPATING",
|
| 196 |
"41": "FAIL",
|
| 197 |
"42": "NOT_ALIVE",
|
| 198 |
"43": "NON_PARTICIPATING",
|
| 199 |
-
"44": "
|
| 200 |
"45": "NOT_ALIVE",
|
| 201 |
-
"46": "
|
| 202 |
"47": "NOT_ALIVE",
|
| 203 |
"48": "NON_PARTICIPATING",
|
| 204 |
"49": "NON_PARTICIPATING",
|
| 205 |
-
"5": "
|
| 206 |
"50": "NOT_ALIVE",
|
| 207 |
-
"51": "
|
| 208 |
"52": "SUCCESS",
|
| 209 |
"53": "SUCCESS",
|
| 210 |
"54": "NON_PARTICIPATING",
|
| 211 |
"55": "NON_PARTICIPATING",
|
| 212 |
-
"56": "
|
| 213 |
"57": "NOT_ALIVE",
|
| 214 |
-
"58": "
|
| 215 |
"59": "NOT_ALIVE",
|
| 216 |
"6": "NOT_ALIVE",
|
| 217 |
"60": "NOT_ALIVE",
|
|
@@ -221,26 +221,26 @@
|
|
| 221 |
"64": "NOT_ALIVE",
|
| 222 |
"65": "NOT_ALIVE",
|
| 223 |
"66": "NON_PARTICIPATING",
|
| 224 |
-
"67": "
|
| 225 |
-
"68": "
|
| 226 |
-
"69": "
|
| 227 |
"7": "NOT_ALIVE",
|
| 228 |
"70": "NOT_ALIVE",
|
| 229 |
-
"71": "
|
| 230 |
"72": "NOT_ALIVE",
|
| 231 |
"73": "NOT_ALIVE",
|
| 232 |
-
"74": "
|
| 233 |
"75": "NOT_ALIVE",
|
| 234 |
"76": "NOT_ALIVE",
|
| 235 |
-
"77": "
|
| 236 |
"78": "NON_PARTICIPATING",
|
| 237 |
"79": "SUCCESS",
|
| 238 |
"8": "NOT_ALIVE",
|
| 239 |
-
"80": "
|
| 240 |
"81": "NOT_ALIVE",
|
| 241 |
"82": "NON_PARTICIPATING",
|
| 242 |
-
"83": "
|
| 243 |
-
"84": "
|
| 244 |
"85": "NOT_ALIVE",
|
| 245 |
"86": "NOT_ALIVE",
|
| 246 |
"87": "FAIL",
|
|
@@ -253,9 +253,9 @@
|
|
| 253 |
"93": "NOT_ALIVE",
|
| 254 |
"94": "NOT_ALIVE",
|
| 255 |
"95": "NOT_ALIVE",
|
| 256 |
-
"96": "
|
| 257 |
"97": "FAIL",
|
| 258 |
-
"98": "
|
| 259 |
"99": "NON_PARTICIPATING"
|
| 260 |
},
|
| 261 |
"architectures": [
|
|
@@ -264,19 +264,17 @@
|
|
| 264 |
"attention_bias": false,
|
| 265 |
"attention_dropout": 0.0,
|
| 266 |
"block_list": [
|
| 267 |
-
|
| 268 |
-
|
| 269 |
-
6373262,
|
| 270 |
-
6373276
|
| 271 |
],
|
| 272 |
"bos_token_id": 1,
|
| 273 |
"eos_token_id": 2,
|
| 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",
|
|
|
|
| 4 |
"0": "NOT_ALIVE",
|
| 5 |
"1": "NOT_ALIVE",
|
| 6 |
"10": "NOT_ALIVE",
|
| 7 |
+
"100": "NOT_ALIVE",
|
| 8 |
"101": "NOT_ALIVE",
|
| 9 |
"102": "NON_PARTICIPATING",
|
| 10 |
"103": "NOT_ALIVE",
|
|
|
|
| 17 |
"11": "NOT_ALIVE",
|
| 18 |
"110": "NON_PARTICIPATING",
|
| 19 |
"111": "NOT_ALIVE",
|
| 20 |
+
"112": "NON_PARTICIPATING",
|
| 21 |
"113": "NOT_ALIVE",
|
| 22 |
"114": "NOT_ALIVE",
|
| 23 |
+
"115": "NON_PARTICIPATING",
|
| 24 |
"116": "NOT_ALIVE",
|
| 25 |
"117": "NOT_ALIVE",
|
| 26 |
"118": "NOT_ALIVE",
|
|
|
|
| 32 |
"123": "SUCCESS",
|
| 33 |
"124": "NOT_ALIVE",
|
| 34 |
"125": "NOT_ALIVE",
|
| 35 |
+
"126": "NON_PARTICIPATING",
|
| 36 |
+
"127": "SUCCESS",
|
| 37 |
"128": "SUCCESS",
|
| 38 |
+
"129": "NOT_ALIVE",
|
| 39 |
"13": "NOT_ALIVE",
|
| 40 |
+
"130": "NOT_ALIVE",
|
| 41 |
"131": "NOT_ALIVE",
|
| 42 |
"132": "SUCCESS",
|
| 43 |
"133": "SUCCESS",
|
| 44 |
+
"134": "NOT_ALIVE",
|
| 45 |
"135": "SUCCESS",
|
| 46 |
"136": "SUCCESS",
|
| 47 |
+
"137": "NOT_ALIVE",
|
| 48 |
"138": "NON_PARTICIPATING",
|
| 49 |
"139": "NOT_ALIVE",
|
| 50 |
"14": "NOT_ALIVE",
|
| 51 |
"140": "NON_PARTICIPATING",
|
| 52 |
"141": "NOT_ALIVE",
|
| 53 |
"142": "NON_PARTICIPATING",
|
| 54 |
+
"143": "NON_PARTICIPATING",
|
| 55 |
"144": "NON_PARTICIPATING",
|
| 56 |
"145": "SUCCESS",
|
| 57 |
"146": "NON_PARTICIPATING",
|
|
|
|
| 59 |
"148": "NON_PARTICIPATING",
|
| 60 |
"149": "SUCCESS",
|
| 61 |
"15": "SUCCESS",
|
| 62 |
+
"150": "NOT_ALIVE",
|
| 63 |
"151": "NON_PARTICIPATING",
|
| 64 |
"152": "FAIL",
|
| 65 |
"153": "SUCCESS",
|
| 66 |
"154": "NOT_ALIVE",
|
| 67 |
+
"155": "SUCCESS",
|
| 68 |
"156": "NOT_ALIVE",
|
| 69 |
"157": "NON_PARTICIPATING",
|
| 70 |
"158": "NON_PARTICIPATING",
|
|
|
|
| 72 |
"16": "NOT_ALIVE",
|
| 73 |
"160": "NON_PARTICIPATING",
|
| 74 |
"161": "NON_PARTICIPATING",
|
| 75 |
+
"162": "NOT_ALIVE",
|
| 76 |
+
"163": "FAIL",
|
| 77 |
"164": "FAIL",
|
| 78 |
"165": "SUCCESS",
|
| 79 |
"166": "NOT_ALIVE",
|
|
|
|
| 86 |
"172": "NON_PARTICIPATING",
|
| 87 |
"173": "NOT_ALIVE",
|
| 88 |
"174": "SUCCESS",
|
| 89 |
+
"175": "NOT_ALIVE",
|
| 90 |
"176": "NOT_ALIVE",
|
| 91 |
"177": "SUCCESS",
|
| 92 |
"178": "NON_PARTICIPATING",
|
|
|
|
| 95 |
"180": "NON_PARTICIPATING",
|
| 96 |
"181": "FAIL",
|
| 97 |
"182": "NOT_ALIVE",
|
| 98 |
+
"183": "NON_PARTICIPATING",
|
| 99 |
"184": "NOT_ALIVE",
|
| 100 |
+
"185": "NOT_ALIVE",
|
| 101 |
"186": "FAIL",
|
| 102 |
"187": "NOT_ALIVE",
|
| 103 |
"188": "FAIL",
|
| 104 |
"189": "SUCCESS",
|
| 105 |
"19": "NOT_ALIVE",
|
| 106 |
+
"190": "SUCCESS",
|
| 107 |
+
"191": "FAIL",
|
| 108 |
+
"192": "FAIL",
|
| 109 |
"193": "NOT_ALIVE",
|
| 110 |
"194": "NON_PARTICIPATING",
|
| 111 |
"195": "NON_PARTICIPATING",
|
| 112 |
"196": "SUCCESS",
|
| 113 |
"197": "NOT_ALIVE",
|
| 114 |
"198": "NOT_ALIVE",
|
| 115 |
+
"199": "NON_PARTICIPATING",
|
| 116 |
"2": "NOT_ALIVE",
|
| 117 |
"20": "NOT_ALIVE",
|
| 118 |
+
"200": "NON_PARTICIPATING",
|
| 119 |
"201": "NOT_ALIVE",
|
| 120 |
"202": "FAIL",
|
| 121 |
"203": "NOT_ALIVE",
|
| 122 |
"204": "NON_PARTICIPATING",
|
| 123 |
"205": "NOT_ALIVE",
|
| 124 |
+
"206": "FAIL",
|
| 125 |
"207": "NOT_ALIVE",
|
| 126 |
"208": "NOT_ALIVE",
|
| 127 |
"209": "NOT_ALIVE",
|
|
|
|
| 137 |
"218": "FAIL",
|
| 138 |
"219": "FAIL",
|
| 139 |
"22": "NOT_ALIVE",
|
| 140 |
+
"220": "NOT_ALIVE",
|
| 141 |
"221": "SUCCESS",
|
| 142 |
"222": "NOT_ALIVE",
|
| 143 |
"223": "NOT_ALIVE",
|
| 144 |
+
"224": "NOT_ALIVE",
|
| 145 |
"225": "SUCCESS",
|
| 146 |
"226": "NOT_ALIVE",
|
| 147 |
"227": "NON_PARTICIPATING",
|
|
|
|
| 149 |
"229": "NON_PARTICIPATING",
|
| 150 |
"23": "NOT_ALIVE",
|
| 151 |
"230": "NOT_ALIVE",
|
| 152 |
+
"231": "NON_PARTICIPATING",
|
| 153 |
"232": "NOT_ALIVE",
|
| 154 |
"233": "NON_PARTICIPATING",
|
| 155 |
"234": "SUCCESS",
|
| 156 |
"235": "NOT_ALIVE",
|
| 157 |
"236": "SUCCESS",
|
| 158 |
"237": "NON_PARTICIPATING",
|
| 159 |
+
"238": "NOT_ALIVE",
|
| 160 |
"239": "NON_PARTICIPATING",
|
| 161 |
"24": "SUCCESS",
|
| 162 |
"240": "NON_PARTICIPATING",
|
|
|
|
| 164 |
"242": "NON_PARTICIPATING",
|
| 165 |
"243": "NON_PARTICIPATING",
|
| 166 |
"244": "NON_PARTICIPATING",
|
| 167 |
+
"245": "NOT_ALIVE",
|
| 168 |
"246": "NON_PARTICIPATING",
|
| 169 |
"247": "NON_PARTICIPATING",
|
| 170 |
"248": "NON_PARTICIPATING",
|
|
|
|
| 172 |
"25": "SUCCESS",
|
| 173 |
"250": "NON_PARTICIPATING",
|
| 174 |
"251": "NON_PARTICIPATING",
|
| 175 |
+
"252": "NOT_ALIVE",
|
| 176 |
"253": "SUCCESS",
|
| 177 |
"254": "NOT_ALIVE",
|
| 178 |
"255": "NON_PARTICIPATING",
|
| 179 |
+
"26": "NOT_ALIVE",
|
| 180 |
"27": "SUCCESS",
|
| 181 |
"28": "NON_PARTICIPATING",
|
| 182 |
"29": "SUCCESS",
|
| 183 |
"3": "FAIL",
|
| 184 |
"30": "SUCCESS",
|
| 185 |
+
"31": "NOT_ALIVE",
|
| 186 |
"32": "NON_PARTICIPATING",
|
| 187 |
"33": "NOT_ALIVE",
|
| 188 |
"34": "NOT_ALIVE",
|
| 189 |
+
"35": "NOT_ALIVE",
|
| 190 |
"36": "FAIL",
|
| 191 |
"37": "NOT_ALIVE",
|
| 192 |
"38": "SUCCESS",
|
| 193 |
+
"39": "NON_PARTICIPATING",
|
| 194 |
"4": "NOT_ALIVE",
|
| 195 |
"40": "NON_PARTICIPATING",
|
| 196 |
"41": "FAIL",
|
| 197 |
"42": "NOT_ALIVE",
|
| 198 |
"43": "NON_PARTICIPATING",
|
| 199 |
+
"44": "NOT_ALIVE",
|
| 200 |
"45": "NOT_ALIVE",
|
| 201 |
+
"46": "SUCCESS",
|
| 202 |
"47": "NOT_ALIVE",
|
| 203 |
"48": "NON_PARTICIPATING",
|
| 204 |
"49": "NON_PARTICIPATING",
|
| 205 |
+
"5": "FAIL",
|
| 206 |
"50": "NOT_ALIVE",
|
| 207 |
+
"51": "NOT_ALIVE",
|
| 208 |
"52": "SUCCESS",
|
| 209 |
"53": "SUCCESS",
|
| 210 |
"54": "NON_PARTICIPATING",
|
| 211 |
"55": "NON_PARTICIPATING",
|
| 212 |
+
"56": "NON_PARTICIPATING",
|
| 213 |
"57": "NOT_ALIVE",
|
| 214 |
+
"58": "SUCCESS",
|
| 215 |
"59": "NOT_ALIVE",
|
| 216 |
"6": "NOT_ALIVE",
|
| 217 |
"60": "NOT_ALIVE",
|
|
|
|
| 221 |
"64": "NOT_ALIVE",
|
| 222 |
"65": "NOT_ALIVE",
|
| 223 |
"66": "NON_PARTICIPATING",
|
| 224 |
+
"67": "NOT_ALIVE",
|
| 225 |
+
"68": "NON_PARTICIPATING",
|
| 226 |
+
"69": "NOT_ALIVE",
|
| 227 |
"7": "NOT_ALIVE",
|
| 228 |
"70": "NOT_ALIVE",
|
| 229 |
+
"71": "NON_PARTICIPATING",
|
| 230 |
"72": "NOT_ALIVE",
|
| 231 |
"73": "NOT_ALIVE",
|
| 232 |
+
"74": "NON_PARTICIPATING",
|
| 233 |
"75": "NOT_ALIVE",
|
| 234 |
"76": "NOT_ALIVE",
|
| 235 |
+
"77": "SUCCESS",
|
| 236 |
"78": "NON_PARTICIPATING",
|
| 237 |
"79": "SUCCESS",
|
| 238 |
"8": "NOT_ALIVE",
|
| 239 |
+
"80": "NOT_ALIVE",
|
| 240 |
"81": "NOT_ALIVE",
|
| 241 |
"82": "NON_PARTICIPATING",
|
| 242 |
+
"83": "NON_PARTICIPATING",
|
| 243 |
+
"84": "FAIL",
|
| 244 |
"85": "NOT_ALIVE",
|
| 245 |
"86": "NOT_ALIVE",
|
| 246 |
"87": "FAIL",
|
|
|
|
| 253 |
"93": "NOT_ALIVE",
|
| 254 |
"94": "NOT_ALIVE",
|
| 255 |
"95": "NOT_ALIVE",
|
| 256 |
+
"96": "NON_PARTICIPATING",
|
| 257 |
"97": "FAIL",
|
| 258 |
+
"98": "FAIL",
|
| 259 |
"99": "NON_PARTICIPATING"
|
| 260 |
},
|
| 261 |
"architectures": [
|
|
|
|
| 264 |
"attention_bias": false,
|
| 265 |
"attention_dropout": 0.0,
|
| 266 |
"block_list": [
|
| 267 |
+
6373462,
|
| 268 |
+
6373476
|
|
|
|
|
|
|
| 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": 6373432,
|
| 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 88389626
|
|
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:343606a790dcef77b4886c1cbbf3ed53445b74a39ac38e4fe81ddd46a01addcb
|
| 3 |
size 88389626
|
inner_optimizer.pt
CHANGED
|
@@ -1,3 +1,3 @@
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
-
oid sha256:
|
| 3 |
size 4924926906
|
|
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:fcfd565d370072865cd72dcaa103f35127ee2071640f98345c3e0240312941a1
|
| 3 |
size 4924926906
|
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:dece5503b60bdf59fcb357ab2423a69e0aa05d71b4686b651a759f52625233dc
|
| 3 |
size 4400216536
|