Run 5. Outer Step 117. Inner Step 2.
Browse files- config.json +54 -54
- inner_optimizer.pt +1 -1
- model.safetensors +1 -1
config.json
CHANGED
|
@@ -30,9 +30,9 @@
|
|
| 30 |
"121": "NOT_ALIVE",
|
| 31 |
"122": "NON_PARTICIPATING",
|
| 32 |
"123": "NOT_ALIVE",
|
| 33 |
-
"124": "
|
| 34 |
-
"125": "
|
| 35 |
-
"126": "
|
| 36 |
"127": "NOT_ALIVE",
|
| 37 |
"128": "NOT_ALIVE",
|
| 38 |
"129": "NON_PARTICIPATING",
|
|
@@ -40,51 +40,51 @@
|
|
| 40 |
"130": "NOT_ALIVE",
|
| 41 |
"131": "NOT_ALIVE",
|
| 42 |
"132": "NON_PARTICIPATING",
|
| 43 |
-
"133": "
|
| 44 |
"134": "NON_PARTICIPATING",
|
| 45 |
"135": "NON_PARTICIPATING",
|
| 46 |
"136": "SUCCESS",
|
| 47 |
"137": "NON_PARTICIPATING",
|
| 48 |
"138": "NOT_ALIVE",
|
| 49 |
"139": "NON_PARTICIPATING",
|
| 50 |
-
"14": "
|
| 51 |
"140": "NON_PARTICIPATING",
|
| 52 |
"141": "NON_PARTICIPATING",
|
| 53 |
"142": "NOT_ALIVE",
|
| 54 |
-
"143": "
|
| 55 |
"144": "NON_PARTICIPATING",
|
| 56 |
"145": "NON_PARTICIPATING",
|
| 57 |
"146": "NON_PARTICIPATING",
|
| 58 |
"147": "NON_PARTICIPATING",
|
| 59 |
-
"148": "
|
| 60 |
"149": "NON_PARTICIPATING",
|
| 61 |
"15": "NOT_ALIVE",
|
| 62 |
-
"150": "
|
| 63 |
"151": "NOT_ALIVE",
|
| 64 |
"152": "NOT_ALIVE",
|
| 65 |
-
"153": "
|
| 66 |
"154": "NON_PARTICIPATING",
|
| 67 |
-
"155": "
|
| 68 |
"156": "NOT_ALIVE",
|
| 69 |
"157": "NON_PARTICIPATING",
|
| 70 |
"158": "NON_PARTICIPATING",
|
| 71 |
-
"159": "
|
| 72 |
"16": "NON_PARTICIPATING",
|
| 73 |
"160": "NON_PARTICIPATING",
|
| 74 |
-
"161": "
|
| 75 |
-
"162": "
|
| 76 |
"163": "NOT_ALIVE",
|
| 77 |
-
"164": "
|
| 78 |
-
"165": "
|
| 79 |
-
"166": "
|
| 80 |
-
"167": "
|
| 81 |
"168": "SUCCESS",
|
| 82 |
"169": "NOT_ALIVE",
|
| 83 |
"17": "NOT_ALIVE",
|
| 84 |
"170": "NON_PARTICIPATING",
|
| 85 |
"171": "NOT_ALIVE",
|
| 86 |
"172": "NOT_ALIVE",
|
| 87 |
-
"173": "
|
| 88 |
"174": "NON_PARTICIPATING",
|
| 89 |
"175": "NOT_ALIVE",
|
| 90 |
"176": "NOT_ALIVE",
|
|
@@ -92,15 +92,15 @@
|
|
| 92 |
"178": "NON_PARTICIPATING",
|
| 93 |
"179": "NOT_ALIVE",
|
| 94 |
"18": "NOT_ALIVE",
|
| 95 |
-
"180": "
|
| 96 |
"181": "NOT_ALIVE",
|
| 97 |
"182": "NOT_ALIVE",
|
| 98 |
"183": "SUCCESS",
|
| 99 |
"184": "NON_PARTICIPATING",
|
| 100 |
"185": "NOT_ALIVE",
|
| 101 |
-
"186": "
|
| 102 |
"187": "NOT_ALIVE",
|
| 103 |
-
"188": "
|
| 104 |
"189": "NOT_ALIVE",
|
| 105 |
"19": "NOT_ALIVE",
|
| 106 |
"190": "NON_PARTICIPATING",
|
|
@@ -109,8 +109,8 @@
|
|
| 109 |
"193": "NON_PARTICIPATING",
|
| 110 |
"194": "NON_PARTICIPATING",
|
| 111 |
"195": "NON_PARTICIPATING",
|
| 112 |
-
"196": "
|
| 113 |
-
"197": "
|
| 114 |
"198": "NOT_ALIVE",
|
| 115 |
"199": "NON_PARTICIPATING",
|
| 116 |
"2": "NOT_ALIVE",
|
|
@@ -121,20 +121,20 @@
|
|
| 121 |
"203": "NOT_ALIVE",
|
| 122 |
"204": "SUCCESS",
|
| 123 |
"205": "NON_PARTICIPATING",
|
| 124 |
-
"206": "
|
| 125 |
"207": "NOT_ALIVE",
|
| 126 |
"208": "NON_PARTICIPATING",
|
| 127 |
"209": "NOT_ALIVE",
|
| 128 |
"21": "NOT_ALIVE",
|
| 129 |
"210": "NON_PARTICIPATING",
|
| 130 |
"211": "NON_PARTICIPATING",
|
| 131 |
-
"212": "
|
| 132 |
"213": "SUCCESS",
|
| 133 |
"214": "NON_PARTICIPATING",
|
| 134 |
-
"215": "
|
| 135 |
"216": "NON_PARTICIPATING",
|
| 136 |
"217": "NON_PARTICIPATING",
|
| 137 |
-
"218": "
|
| 138 |
"219": "NOT_ALIVE",
|
| 139 |
"22": "NOT_ALIVE",
|
| 140 |
"220": "NON_PARTICIPATING",
|
|
@@ -149,30 +149,30 @@
|
|
| 149 |
"229": "NON_PARTICIPATING",
|
| 150 |
"23": "NON_PARTICIPATING",
|
| 151 |
"230": "NON_PARTICIPATING",
|
| 152 |
-
"231": "
|
| 153 |
"232": "NOT_ALIVE",
|
| 154 |
"233": "NOT_ALIVE",
|
| 155 |
"234": "NOT_ALIVE",
|
| 156 |
-
"235": "
|
| 157 |
"236": "NOT_ALIVE",
|
| 158 |
-
"237": "
|
| 159 |
-
"238": "
|
| 160 |
"239": "NOT_ALIVE",
|
| 161 |
"24": "NON_PARTICIPATING",
|
| 162 |
"240": "NOT_ALIVE",
|
| 163 |
-
"241": "
|
| 164 |
"242": "NOT_ALIVE",
|
| 165 |
"243": "NOT_ALIVE",
|
| 166 |
"244": "NOT_ALIVE",
|
| 167 |
-
"245": "
|
| 168 |
-
"246": "
|
| 169 |
"247": "NON_PARTICIPATING",
|
| 170 |
"248": "NOT_ALIVE",
|
| 171 |
"249": "NOT_ALIVE",
|
| 172 |
"25": "SUCCESS",
|
| 173 |
-
"250": "
|
| 174 |
"251": "NOT_ALIVE",
|
| 175 |
-
"252": "
|
| 176 |
"253": "NOT_ALIVE",
|
| 177 |
"254": "NON_PARTICIPATING",
|
| 178 |
"255": "NON_PARTICIPATING",
|
|
@@ -181,12 +181,12 @@
|
|
| 181 |
"28": "NOT_ALIVE",
|
| 182 |
"29": "NOT_ALIVE",
|
| 183 |
"3": "NON_PARTICIPATING",
|
| 184 |
-
"30": "
|
| 185 |
"31": "NON_PARTICIPATING",
|
| 186 |
-
"32": "
|
| 187 |
"33": "NOT_ALIVE",
|
| 188 |
"34": "NON_PARTICIPATING",
|
| 189 |
-
"35": "
|
| 190 |
"36": "NOT_ALIVE",
|
| 191 |
"37": "NOT_ALIVE",
|
| 192 |
"38": "SUCCESS",
|
|
@@ -195,29 +195,29 @@
|
|
| 195 |
"40": "NON_PARTICIPATING",
|
| 196 |
"41": "NON_PARTICIPATING",
|
| 197 |
"42": "NON_PARTICIPATING",
|
| 198 |
-
"43": "
|
| 199 |
"44": "NON_PARTICIPATING",
|
| 200 |
"45": "NOT_ALIVE",
|
| 201 |
"46": "NOT_ALIVE",
|
| 202 |
"47": "NON_PARTICIPATING",
|
| 203 |
-
"48": "
|
| 204 |
"49": "NON_PARTICIPATING",
|
| 205 |
"5": "SUCCESS",
|
| 206 |
"50": "NOT_ALIVE",
|
| 207 |
"51": "NON_PARTICIPATING",
|
| 208 |
-
"52": "
|
| 209 |
"53": "NON_PARTICIPATING",
|
| 210 |
"54": "NOT_ALIVE",
|
| 211 |
"55": "NOT_ALIVE",
|
| 212 |
"56": "NON_PARTICIPATING",
|
| 213 |
"57": "NON_PARTICIPATING",
|
| 214 |
-
"58": "
|
| 215 |
"59": "SUCCESS",
|
| 216 |
"6": "NON_PARTICIPATING",
|
| 217 |
"60": "NON_PARTICIPATING",
|
| 218 |
"61": "NON_PARTICIPATING",
|
| 219 |
"62": "NOT_ALIVE",
|
| 220 |
-
"63": "
|
| 221 |
"64": "NOT_ALIVE",
|
| 222 |
"65": "NON_PARTICIPATING",
|
| 223 |
"66": "NOT_ALIVE",
|
|
@@ -228,7 +228,7 @@
|
|
| 228 |
"70": "NON_PARTICIPATING",
|
| 229 |
"71": "NON_PARTICIPATING",
|
| 230 |
"72": "NOT_ALIVE",
|
| 231 |
-
"73": "
|
| 232 |
"74": "NON_PARTICIPATING",
|
| 233 |
"75": "NON_PARTICIPATING",
|
| 234 |
"76": "NON_PARTICIPATING",
|
|
@@ -246,12 +246,12 @@
|
|
| 246 |
"87": "NOT_ALIVE",
|
| 247 |
"88": "NON_PARTICIPATING",
|
| 248 |
"89": "NOT_ALIVE",
|
| 249 |
-
"9": "
|
| 250 |
-
"90": "
|
| 251 |
-
"91": "
|
| 252 |
-
"92": "
|
| 253 |
"93": "NON_PARTICIPATING",
|
| 254 |
-
"94": "
|
| 255 |
"95": "NOT_ALIVE",
|
| 256 |
"96": "NON_PARTICIPATING",
|
| 257 |
"97": "NON_PARTICIPATING",
|
|
@@ -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",
|
|
|
|
| 30 |
"121": "NOT_ALIVE",
|
| 31 |
"122": "NON_PARTICIPATING",
|
| 32 |
"123": "NOT_ALIVE",
|
| 33 |
+
"124": "NON_PARTICIPATING",
|
| 34 |
+
"125": "NON_PARTICIPATING",
|
| 35 |
+
"126": "NON_PARTICIPATING",
|
| 36 |
"127": "NOT_ALIVE",
|
| 37 |
"128": "NOT_ALIVE",
|
| 38 |
"129": "NON_PARTICIPATING",
|
|
|
|
| 40 |
"130": "NOT_ALIVE",
|
| 41 |
"131": "NOT_ALIVE",
|
| 42 |
"132": "NON_PARTICIPATING",
|
| 43 |
+
"133": "NON_PARTICIPATING",
|
| 44 |
"134": "NON_PARTICIPATING",
|
| 45 |
"135": "NON_PARTICIPATING",
|
| 46 |
"136": "SUCCESS",
|
| 47 |
"137": "NON_PARTICIPATING",
|
| 48 |
"138": "NOT_ALIVE",
|
| 49 |
"139": "NON_PARTICIPATING",
|
| 50 |
+
"14": "NON_PARTICIPATING",
|
| 51 |
"140": "NON_PARTICIPATING",
|
| 52 |
"141": "NON_PARTICIPATING",
|
| 53 |
"142": "NOT_ALIVE",
|
| 54 |
+
"143": "NON_PARTICIPATING",
|
| 55 |
"144": "NON_PARTICIPATING",
|
| 56 |
"145": "NON_PARTICIPATING",
|
| 57 |
"146": "NON_PARTICIPATING",
|
| 58 |
"147": "NON_PARTICIPATING",
|
| 59 |
+
"148": "SUCCESS",
|
| 60 |
"149": "NON_PARTICIPATING",
|
| 61 |
"15": "NOT_ALIVE",
|
| 62 |
+
"150": "NON_PARTICIPATING",
|
| 63 |
"151": "NOT_ALIVE",
|
| 64 |
"152": "NOT_ALIVE",
|
| 65 |
+
"153": "SUCCESS",
|
| 66 |
"154": "NON_PARTICIPATING",
|
| 67 |
+
"155": "NON_PARTICIPATING",
|
| 68 |
"156": "NOT_ALIVE",
|
| 69 |
"157": "NON_PARTICIPATING",
|
| 70 |
"158": "NON_PARTICIPATING",
|
| 71 |
+
"159": "NOT_ALIVE",
|
| 72 |
"16": "NON_PARTICIPATING",
|
| 73 |
"160": "NON_PARTICIPATING",
|
| 74 |
+
"161": "NOT_ALIVE",
|
| 75 |
+
"162": "NON_PARTICIPATING",
|
| 76 |
"163": "NOT_ALIVE",
|
| 77 |
+
"164": "NON_PARTICIPATING",
|
| 78 |
+
"165": "SUCCESS",
|
| 79 |
+
"166": "NOT_ALIVE",
|
| 80 |
+
"167": "NON_PARTICIPATING",
|
| 81 |
"168": "SUCCESS",
|
| 82 |
"169": "NOT_ALIVE",
|
| 83 |
"17": "NOT_ALIVE",
|
| 84 |
"170": "NON_PARTICIPATING",
|
| 85 |
"171": "NOT_ALIVE",
|
| 86 |
"172": "NOT_ALIVE",
|
| 87 |
+
"173": "SUCCESS",
|
| 88 |
"174": "NON_PARTICIPATING",
|
| 89 |
"175": "NOT_ALIVE",
|
| 90 |
"176": "NOT_ALIVE",
|
|
|
|
| 92 |
"178": "NON_PARTICIPATING",
|
| 93 |
"179": "NOT_ALIVE",
|
| 94 |
"18": "NOT_ALIVE",
|
| 95 |
+
"180": "NOT_ALIVE",
|
| 96 |
"181": "NOT_ALIVE",
|
| 97 |
"182": "NOT_ALIVE",
|
| 98 |
"183": "SUCCESS",
|
| 99 |
"184": "NON_PARTICIPATING",
|
| 100 |
"185": "NOT_ALIVE",
|
| 101 |
+
"186": "NON_PARTICIPATING",
|
| 102 |
"187": "NOT_ALIVE",
|
| 103 |
+
"188": "NOT_ALIVE",
|
| 104 |
"189": "NOT_ALIVE",
|
| 105 |
"19": "NOT_ALIVE",
|
| 106 |
"190": "NON_PARTICIPATING",
|
|
|
|
| 109 |
"193": "NON_PARTICIPATING",
|
| 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",
|
|
|
|
| 121 |
"203": "NOT_ALIVE",
|
| 122 |
"204": "SUCCESS",
|
| 123 |
"205": "NON_PARTICIPATING",
|
| 124 |
+
"206": "NON_PARTICIPATING",
|
| 125 |
"207": "NOT_ALIVE",
|
| 126 |
"208": "NON_PARTICIPATING",
|
| 127 |
"209": "NOT_ALIVE",
|
| 128 |
"21": "NOT_ALIVE",
|
| 129 |
"210": "NON_PARTICIPATING",
|
| 130 |
"211": "NON_PARTICIPATING",
|
| 131 |
+
"212": "NON_PARTICIPATING",
|
| 132 |
"213": "SUCCESS",
|
| 133 |
"214": "NON_PARTICIPATING",
|
| 134 |
+
"215": "NON_PARTICIPATING",
|
| 135 |
"216": "NON_PARTICIPATING",
|
| 136 |
"217": "NON_PARTICIPATING",
|
| 137 |
+
"218": "NOT_ALIVE",
|
| 138 |
"219": "NOT_ALIVE",
|
| 139 |
"22": "NOT_ALIVE",
|
| 140 |
"220": "NON_PARTICIPATING",
|
|
|
|
| 149 |
"229": "NON_PARTICIPATING",
|
| 150 |
"23": "NON_PARTICIPATING",
|
| 151 |
"230": "NON_PARTICIPATING",
|
| 152 |
+
"231": "NON_PARTICIPATING",
|
| 153 |
"232": "NOT_ALIVE",
|
| 154 |
"233": "NOT_ALIVE",
|
| 155 |
"234": "NOT_ALIVE",
|
| 156 |
+
"235": "NON_PARTICIPATING",
|
| 157 |
"236": "NOT_ALIVE",
|
| 158 |
+
"237": "NOT_ALIVE",
|
| 159 |
+
"238": "SUCCESS",
|
| 160 |
"239": "NOT_ALIVE",
|
| 161 |
"24": "NON_PARTICIPATING",
|
| 162 |
"240": "NOT_ALIVE",
|
| 163 |
+
"241": "NON_PARTICIPATING",
|
| 164 |
"242": "NOT_ALIVE",
|
| 165 |
"243": "NOT_ALIVE",
|
| 166 |
"244": "NOT_ALIVE",
|
| 167 |
+
"245": "NOT_ALIVE",
|
| 168 |
+
"246": "NOT_ALIVE",
|
| 169 |
"247": "NON_PARTICIPATING",
|
| 170 |
"248": "NOT_ALIVE",
|
| 171 |
"249": "NOT_ALIVE",
|
| 172 |
"25": "SUCCESS",
|
| 173 |
+
"250": "NOT_ALIVE",
|
| 174 |
"251": "NOT_ALIVE",
|
| 175 |
+
"252": "NON_PARTICIPATING",
|
| 176 |
"253": "NOT_ALIVE",
|
| 177 |
"254": "NON_PARTICIPATING",
|
| 178 |
"255": "NON_PARTICIPATING",
|
|
|
|
| 181 |
"28": "NOT_ALIVE",
|
| 182 |
"29": "NOT_ALIVE",
|
| 183 |
"3": "NON_PARTICIPATING",
|
| 184 |
+
"30": "SUCCESS",
|
| 185 |
"31": "NON_PARTICIPATING",
|
| 186 |
+
"32": "SUCCESS",
|
| 187 |
"33": "NOT_ALIVE",
|
| 188 |
"34": "NON_PARTICIPATING",
|
| 189 |
+
"35": "SUCCESS",
|
| 190 |
"36": "NOT_ALIVE",
|
| 191 |
"37": "NOT_ALIVE",
|
| 192 |
"38": "SUCCESS",
|
|
|
|
| 195 |
"40": "NON_PARTICIPATING",
|
| 196 |
"41": "NON_PARTICIPATING",
|
| 197 |
"42": "NON_PARTICIPATING",
|
| 198 |
+
"43": "NOT_ALIVE",
|
| 199 |
"44": "NON_PARTICIPATING",
|
| 200 |
"45": "NOT_ALIVE",
|
| 201 |
"46": "NOT_ALIVE",
|
| 202 |
"47": "NON_PARTICIPATING",
|
| 203 |
+
"48": "NON_PARTICIPATING",
|
| 204 |
"49": "NON_PARTICIPATING",
|
| 205 |
"5": "SUCCESS",
|
| 206 |
"50": "NOT_ALIVE",
|
| 207 |
"51": "NON_PARTICIPATING",
|
| 208 |
+
"52": "NON_PARTICIPATING",
|
| 209 |
"53": "NON_PARTICIPATING",
|
| 210 |
"54": "NOT_ALIVE",
|
| 211 |
"55": "NOT_ALIVE",
|
| 212 |
"56": "NON_PARTICIPATING",
|
| 213 |
"57": "NON_PARTICIPATING",
|
| 214 |
+
"58": "NOT_ALIVE",
|
| 215 |
"59": "SUCCESS",
|
| 216 |
"6": "NON_PARTICIPATING",
|
| 217 |
"60": "NON_PARTICIPATING",
|
| 218 |
"61": "NON_PARTICIPATING",
|
| 219 |
"62": "NOT_ALIVE",
|
| 220 |
+
"63": "NON_PARTICIPATING",
|
| 221 |
"64": "NOT_ALIVE",
|
| 222 |
"65": "NON_PARTICIPATING",
|
| 223 |
"66": "NOT_ALIVE",
|
|
|
|
| 228 |
"70": "NON_PARTICIPATING",
|
| 229 |
"71": "NON_PARTICIPATING",
|
| 230 |
"72": "NOT_ALIVE",
|
| 231 |
+
"73": "NOT_ALIVE",
|
| 232 |
"74": "NON_PARTICIPATING",
|
| 233 |
"75": "NON_PARTICIPATING",
|
| 234 |
"76": "NON_PARTICIPATING",
|
|
|
|
| 246 |
"87": "NOT_ALIVE",
|
| 247 |
"88": "NON_PARTICIPATING",
|
| 248 |
"89": "NOT_ALIVE",
|
| 249 |
+
"9": "NON_PARTICIPATING",
|
| 250 |
+
"90": "NOT_ALIVE",
|
| 251 |
+
"91": "NOT_ALIVE",
|
| 252 |
+
"92": "SUCCESS",
|
| 253 |
"93": "NON_PARTICIPATING",
|
| 254 |
+
"94": "NON_PARTICIPATING",
|
| 255 |
"95": "NOT_ALIVE",
|
| 256 |
"96": "NON_PARTICIPATING",
|
| 257 |
"97": "NON_PARTICIPATING",
|
|
|
|
| 264 |
"attention_bias": false,
|
| 265 |
"attention_dropout": 0.0,
|
| 266 |
"block_list": [
|
| 267 |
+
5978118,
|
| 268 |
+
5978145
|
| 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": 5978024,
|
| 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:b3f031845cd0bba01b2ac1216b5e3f6f24814e45ae0b624fbb52258c05c4e8ab
|
| 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 sha256:961cfa0a100c59890f2725a0be5f037f57f64d6ca6060f997a959177eaf71382
|
| 3 |
size 4400216536
|