Run 5. Outer Step 67. Inner Step 2.
Browse files- config.json +42 -42
- inner_optimizer.pt +1 -1
- model.safetensors +1 -1
config.json
CHANGED
|
@@ -5,9 +5,9 @@
|
|
| 5 |
"1": "NON_PARTICIPATING",
|
| 6 |
"10": "SUCCESS",
|
| 7 |
"100": "NOT_ALIVE",
|
| 8 |
-
"101": "
|
| 9 |
"102": "NON_PARTICIPATING",
|
| 10 |
-
"103": "
|
| 11 |
"104": "SUCCESS",
|
| 12 |
"105": "NON_PARTICIPATING",
|
| 13 |
"106": "NOT_ALIVE",
|
|
@@ -21,7 +21,7 @@
|
|
| 21 |
"113": "NOT_ALIVE",
|
| 22 |
"114": "NON_PARTICIPATING",
|
| 23 |
"115": "NON_PARTICIPATING",
|
| 24 |
-
"116": "
|
| 25 |
"117": "NON_PARTICIPATING",
|
| 26 |
"118": "NON_PARTICIPATING",
|
| 27 |
"119": "NOT_ALIVE",
|
|
@@ -40,7 +40,7 @@
|
|
| 40 |
"130": "NOT_ALIVE",
|
| 41 |
"131": "NOT_ALIVE",
|
| 42 |
"132": "NOT_ALIVE",
|
| 43 |
-
"133": "
|
| 44 |
"134": "NON_PARTICIPATING",
|
| 45 |
"135": "NON_PARTICIPATING",
|
| 46 |
"136": "NOT_ALIVE",
|
|
@@ -51,10 +51,10 @@
|
|
| 51 |
"140": "NON_PARTICIPATING",
|
| 52 |
"141": "NON_PARTICIPATING",
|
| 53 |
"142": "SUCCESS",
|
| 54 |
-
"143": "
|
| 55 |
"144": "NON_PARTICIPATING",
|
| 56 |
"145": "NOT_ALIVE",
|
| 57 |
-
"146": "
|
| 58 |
"147": "NON_PARTICIPATING",
|
| 59 |
"148": "NON_PARTICIPATING",
|
| 60 |
"149": "NON_PARTICIPATING",
|
|
@@ -64,20 +64,20 @@
|
|
| 64 |
"152": "NOT_ALIVE",
|
| 65 |
"153": "NOT_ALIVE",
|
| 66 |
"154": "NOT_ALIVE",
|
| 67 |
-
"155": "
|
| 68 |
"156": "NOT_ALIVE",
|
| 69 |
"157": "NOT_ALIVE",
|
| 70 |
-
"158": "
|
| 71 |
"159": "NOT_ALIVE",
|
| 72 |
"16": "NON_PARTICIPATING",
|
| 73 |
-
"160": "
|
| 74 |
"161": "NOT_ALIVE",
|
| 75 |
"162": "NOT_ALIVE",
|
| 76 |
"163": "NOT_ALIVE",
|
| 77 |
-
"164": "
|
| 78 |
"165": "NOT_ALIVE",
|
| 79 |
"166": "NON_PARTICIPATING",
|
| 80 |
-
"167": "
|
| 81 |
"168": "NOT_ALIVE",
|
| 82 |
"169": "NOT_ALIVE",
|
| 83 |
"17": "NOT_ALIVE",
|
|
@@ -85,20 +85,20 @@
|
|
| 85 |
"171": "NON_PARTICIPATING",
|
| 86 |
"172": "NOT_ALIVE",
|
| 87 |
"173": "NON_PARTICIPATING",
|
| 88 |
-
"174": "
|
| 89 |
"175": "NON_PARTICIPATING",
|
| 90 |
"176": "NOT_ALIVE",
|
| 91 |
"177": "NON_PARTICIPATING",
|
| 92 |
"178": "NON_PARTICIPATING",
|
| 93 |
-
"179": "
|
| 94 |
"18": "NON_PARTICIPATING",
|
| 95 |
-
"180": "
|
| 96 |
-
"181": "
|
| 97 |
"182": "NON_PARTICIPATING",
|
| 98 |
"183": "SUCCESS",
|
| 99 |
"184": "NON_PARTICIPATING",
|
| 100 |
"185": "NON_PARTICIPATING",
|
| 101 |
-
"186": "
|
| 102 |
"187": "SUCCESS",
|
| 103 |
"188": "NON_PARTICIPATING",
|
| 104 |
"189": "NON_PARTICIPATING",
|
|
@@ -111,7 +111,7 @@
|
|
| 111 |
"195": "NOT_ALIVE",
|
| 112 |
"196": "NOT_ALIVE",
|
| 113 |
"197": "NOT_ALIVE",
|
| 114 |
-
"198": "
|
| 115 |
"199": "NOT_ALIVE",
|
| 116 |
"2": "NOT_ALIVE",
|
| 117 |
"20": "NOT_ALIVE",
|
|
@@ -120,7 +120,7 @@
|
|
| 120 |
"202": "NON_PARTICIPATING",
|
| 121 |
"203": "NOT_ALIVE",
|
| 122 |
"204": "NON_PARTICIPATING",
|
| 123 |
-
"205": "
|
| 124 |
"206": "NOT_ALIVE",
|
| 125 |
"207": "NOT_ALIVE",
|
| 126 |
"208": "NOT_ALIVE",
|
|
@@ -173,7 +173,7 @@
|
|
| 173 |
"250": "NON_PARTICIPATING",
|
| 174 |
"251": "NOT_ALIVE",
|
| 175 |
"252": "NOT_ALIVE",
|
| 176 |
-
"253": "
|
| 177 |
"254": "NOT_ALIVE",
|
| 178 |
"255": "NON_PARTICIPATING",
|
| 179 |
"26": "NON_PARTICIPATING",
|
|
@@ -181,22 +181,22 @@
|
|
| 181 |
"28": "SUCCESS",
|
| 182 |
"29": "NON_PARTICIPATING",
|
| 183 |
"3": "SUCCESS",
|
| 184 |
-
"30": "
|
| 185 |
-
"31": "
|
| 186 |
-
"32": "
|
| 187 |
"33": "NOT_ALIVE",
|
| 188 |
"34": "NON_PARTICIPATING",
|
| 189 |
"35": "NON_PARTICIPATING",
|
| 190 |
"36": "NON_PARTICIPATING",
|
| 191 |
"37": "NON_PARTICIPATING",
|
| 192 |
-
"38": "
|
| 193 |
"39": "NOT_ALIVE",
|
| 194 |
"4": "NON_PARTICIPATING",
|
| 195 |
"40": "NON_PARTICIPATING",
|
| 196 |
"41": "SUCCESS",
|
| 197 |
"42": "SUCCESS",
|
| 198 |
-
"43": "
|
| 199 |
-
"44": "
|
| 200 |
"45": "NOT_ALIVE",
|
| 201 |
"46": "SUCCESS",
|
| 202 |
"47": "NON_PARTICIPATING",
|
|
@@ -204,32 +204,32 @@
|
|
| 204 |
"49": "NON_PARTICIPATING",
|
| 205 |
"5": "NON_PARTICIPATING",
|
| 206 |
"50": "NOT_ALIVE",
|
| 207 |
-
"51": "
|
| 208 |
"52": "NOT_ALIVE",
|
| 209 |
"53": "NOT_ALIVE",
|
| 210 |
"54": "NON_PARTICIPATING",
|
| 211 |
-
"55": "
|
| 212 |
"56": "NON_PARTICIPATING",
|
| 213 |
"57": "SUCCESS",
|
| 214 |
-
"58": "
|
| 215 |
"59": "NON_PARTICIPATING",
|
| 216 |
"6": "NOT_ALIVE",
|
| 217 |
-
"60": "
|
| 218 |
"61": "NOT_ALIVE",
|
| 219 |
"62": "NON_PARTICIPATING",
|
| 220 |
"63": "NOT_ALIVE",
|
| 221 |
"64": "NON_PARTICIPATING",
|
| 222 |
-
"65": "
|
| 223 |
"66": "NON_PARTICIPATING",
|
| 224 |
"67": "NOT_ALIVE",
|
| 225 |
"68": "NON_PARTICIPATING",
|
| 226 |
-
"69": "
|
| 227 |
"7": "SUCCESS",
|
| 228 |
-
"70": "
|
| 229 |
"71": "NON_PARTICIPATING",
|
| 230 |
"72": "NON_PARTICIPATING",
|
| 231 |
"73": "SUCCESS",
|
| 232 |
-
"74": "
|
| 233 |
"75": "NON_PARTICIPATING",
|
| 234 |
"76": "NON_PARTICIPATING",
|
| 235 |
"77": "NOT_ALIVE",
|
|
@@ -238,21 +238,21 @@
|
|
| 238 |
"8": "SUCCESS",
|
| 239 |
"80": "NON_PARTICIPATING",
|
| 240 |
"81": "NON_PARTICIPATING",
|
| 241 |
-
"82": "
|
| 242 |
-
"83": "
|
| 243 |
"84": "NOT_ALIVE",
|
| 244 |
"85": "NON_PARTICIPATING",
|
| 245 |
-
"86": "
|
| 246 |
"87": "NOT_ALIVE",
|
| 247 |
"88": "NON_PARTICIPATING",
|
| 248 |
"89": "NON_PARTICIPATING",
|
| 249 |
-
"9": "
|
| 250 |
"90": "NON_PARTICIPATING",
|
| 251 |
"91": "NOT_ALIVE",
|
| 252 |
"92": "NON_PARTICIPATING",
|
| 253 |
"93": "NON_PARTICIPATING",
|
| 254 |
"94": "NON_PARTICIPATING",
|
| 255 |
-
"95": "
|
| 256 |
"96": "NON_PARTICIPATING",
|
| 257 |
"97": "NON_PARTICIPATING",
|
| 258 |
"98": "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",
|
|
|
|
| 5 |
"1": "NON_PARTICIPATING",
|
| 6 |
"10": "SUCCESS",
|
| 7 |
"100": "NOT_ALIVE",
|
| 8 |
+
"101": "NOT_ALIVE",
|
| 9 |
"102": "NON_PARTICIPATING",
|
| 10 |
+
"103": "NOT_ALIVE",
|
| 11 |
"104": "SUCCESS",
|
| 12 |
"105": "NON_PARTICIPATING",
|
| 13 |
"106": "NOT_ALIVE",
|
|
|
|
| 21 |
"113": "NOT_ALIVE",
|
| 22 |
"114": "NON_PARTICIPATING",
|
| 23 |
"115": "NON_PARTICIPATING",
|
| 24 |
+
"116": "NOT_ALIVE",
|
| 25 |
"117": "NON_PARTICIPATING",
|
| 26 |
"118": "NON_PARTICIPATING",
|
| 27 |
"119": "NOT_ALIVE",
|
|
|
|
| 40 |
"130": "NOT_ALIVE",
|
| 41 |
"131": "NOT_ALIVE",
|
| 42 |
"132": "NOT_ALIVE",
|
| 43 |
+
"133": "SUCCESS",
|
| 44 |
"134": "NON_PARTICIPATING",
|
| 45 |
"135": "NON_PARTICIPATING",
|
| 46 |
"136": "NOT_ALIVE",
|
|
|
|
| 51 |
"140": "NON_PARTICIPATING",
|
| 52 |
"141": "NON_PARTICIPATING",
|
| 53 |
"142": "SUCCESS",
|
| 54 |
+
"143": "NON_PARTICIPATING",
|
| 55 |
"144": "NON_PARTICIPATING",
|
| 56 |
"145": "NOT_ALIVE",
|
| 57 |
+
"146": "NOT_ALIVE",
|
| 58 |
"147": "NON_PARTICIPATING",
|
| 59 |
"148": "NON_PARTICIPATING",
|
| 60 |
"149": "NON_PARTICIPATING",
|
|
|
|
| 64 |
"152": "NOT_ALIVE",
|
| 65 |
"153": "NOT_ALIVE",
|
| 66 |
"154": "NOT_ALIVE",
|
| 67 |
+
"155": "NOT_ALIVE",
|
| 68 |
"156": "NOT_ALIVE",
|
| 69 |
"157": "NOT_ALIVE",
|
| 70 |
+
"158": "NOT_ALIVE",
|
| 71 |
"159": "NOT_ALIVE",
|
| 72 |
"16": "NON_PARTICIPATING",
|
| 73 |
+
"160": "NON_PARTICIPATING",
|
| 74 |
"161": "NOT_ALIVE",
|
| 75 |
"162": "NOT_ALIVE",
|
| 76 |
"163": "NOT_ALIVE",
|
| 77 |
+
"164": "NON_PARTICIPATING",
|
| 78 |
"165": "NOT_ALIVE",
|
| 79 |
"166": "NON_PARTICIPATING",
|
| 80 |
+
"167": "NOT_ALIVE",
|
| 81 |
"168": "NOT_ALIVE",
|
| 82 |
"169": "NOT_ALIVE",
|
| 83 |
"17": "NOT_ALIVE",
|
|
|
|
| 85 |
"171": "NON_PARTICIPATING",
|
| 86 |
"172": "NOT_ALIVE",
|
| 87 |
"173": "NON_PARTICIPATING",
|
| 88 |
+
"174": "NOT_ALIVE",
|
| 89 |
"175": "NON_PARTICIPATING",
|
| 90 |
"176": "NOT_ALIVE",
|
| 91 |
"177": "NON_PARTICIPATING",
|
| 92 |
"178": "NON_PARTICIPATING",
|
| 93 |
+
"179": "NOT_ALIVE",
|
| 94 |
"18": "NON_PARTICIPATING",
|
| 95 |
+
"180": "NOT_ALIVE",
|
| 96 |
+
"181": "SUCCESS",
|
| 97 |
"182": "NON_PARTICIPATING",
|
| 98 |
"183": "SUCCESS",
|
| 99 |
"184": "NON_PARTICIPATING",
|
| 100 |
"185": "NON_PARTICIPATING",
|
| 101 |
+
"186": "NOT_ALIVE",
|
| 102 |
"187": "SUCCESS",
|
| 103 |
"188": "NON_PARTICIPATING",
|
| 104 |
"189": "NON_PARTICIPATING",
|
|
|
|
| 111 |
"195": "NOT_ALIVE",
|
| 112 |
"196": "NOT_ALIVE",
|
| 113 |
"197": "NOT_ALIVE",
|
| 114 |
+
"198": "NON_PARTICIPATING",
|
| 115 |
"199": "NOT_ALIVE",
|
| 116 |
"2": "NOT_ALIVE",
|
| 117 |
"20": "NOT_ALIVE",
|
|
|
|
| 120 |
"202": "NON_PARTICIPATING",
|
| 121 |
"203": "NOT_ALIVE",
|
| 122 |
"204": "NON_PARTICIPATING",
|
| 123 |
+
"205": "NON_PARTICIPATING",
|
| 124 |
"206": "NOT_ALIVE",
|
| 125 |
"207": "NOT_ALIVE",
|
| 126 |
"208": "NOT_ALIVE",
|
|
|
|
| 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": "NON_PARTICIPATING",
|
|
|
|
| 181 |
"28": "SUCCESS",
|
| 182 |
"29": "NON_PARTICIPATING",
|
| 183 |
"3": "SUCCESS",
|
| 184 |
+
"30": "NOT_ALIVE",
|
| 185 |
+
"31": "NOT_ALIVE",
|
| 186 |
+
"32": "NOT_ALIVE",
|
| 187 |
"33": "NOT_ALIVE",
|
| 188 |
"34": "NON_PARTICIPATING",
|
| 189 |
"35": "NON_PARTICIPATING",
|
| 190 |
"36": "NON_PARTICIPATING",
|
| 191 |
"37": "NON_PARTICIPATING",
|
| 192 |
+
"38": "SUCCESS",
|
| 193 |
"39": "NOT_ALIVE",
|
| 194 |
"4": "NON_PARTICIPATING",
|
| 195 |
"40": "NON_PARTICIPATING",
|
| 196 |
"41": "SUCCESS",
|
| 197 |
"42": "SUCCESS",
|
| 198 |
+
"43": "NOT_ALIVE",
|
| 199 |
+
"44": "NON_PARTICIPATING",
|
| 200 |
"45": "NOT_ALIVE",
|
| 201 |
"46": "SUCCESS",
|
| 202 |
"47": "NON_PARTICIPATING",
|
|
|
|
| 204 |
"49": "NON_PARTICIPATING",
|
| 205 |
"5": "NON_PARTICIPATING",
|
| 206 |
"50": "NOT_ALIVE",
|
| 207 |
+
"51": "NOT_ALIVE",
|
| 208 |
"52": "NOT_ALIVE",
|
| 209 |
"53": "NOT_ALIVE",
|
| 210 |
"54": "NON_PARTICIPATING",
|
| 211 |
+
"55": "NOT_ALIVE",
|
| 212 |
"56": "NON_PARTICIPATING",
|
| 213 |
"57": "SUCCESS",
|
| 214 |
+
"58": "NON_PARTICIPATING",
|
| 215 |
"59": "NON_PARTICIPATING",
|
| 216 |
"6": "NOT_ALIVE",
|
| 217 |
+
"60": "NOT_ALIVE",
|
| 218 |
"61": "NOT_ALIVE",
|
| 219 |
"62": "NON_PARTICIPATING",
|
| 220 |
"63": "NOT_ALIVE",
|
| 221 |
"64": "NON_PARTICIPATING",
|
| 222 |
+
"65": "NOT_ALIVE",
|
| 223 |
"66": "NON_PARTICIPATING",
|
| 224 |
"67": "NOT_ALIVE",
|
| 225 |
"68": "NON_PARTICIPATING",
|
| 226 |
+
"69": "NOT_ALIVE",
|
| 227 |
"7": "SUCCESS",
|
| 228 |
+
"70": "NON_PARTICIPATING",
|
| 229 |
"71": "NON_PARTICIPATING",
|
| 230 |
"72": "NON_PARTICIPATING",
|
| 231 |
"73": "SUCCESS",
|
| 232 |
+
"74": "NON_PARTICIPATING",
|
| 233 |
"75": "NON_PARTICIPATING",
|
| 234 |
"76": "NON_PARTICIPATING",
|
| 235 |
"77": "NOT_ALIVE",
|
|
|
|
| 238 |
"8": "SUCCESS",
|
| 239 |
"80": "NON_PARTICIPATING",
|
| 240 |
"81": "NON_PARTICIPATING",
|
| 241 |
+
"82": "NOT_ALIVE",
|
| 242 |
+
"83": "NOT_ALIVE",
|
| 243 |
"84": "NOT_ALIVE",
|
| 244 |
"85": "NON_PARTICIPATING",
|
| 245 |
+
"86": "NON_PARTICIPATING",
|
| 246 |
"87": "NOT_ALIVE",
|
| 247 |
"88": "NON_PARTICIPATING",
|
| 248 |
"89": "NON_PARTICIPATING",
|
| 249 |
+
"9": "NOT_ALIVE",
|
| 250 |
"90": "NON_PARTICIPATING",
|
| 251 |
"91": "NOT_ALIVE",
|
| 252 |
"92": "NON_PARTICIPATING",
|
| 253 |
"93": "NON_PARTICIPATING",
|
| 254 |
"94": "NON_PARTICIPATING",
|
| 255 |
+
"95": "NOT_ALIVE",
|
| 256 |
"96": "NON_PARTICIPATING",
|
| 257 |
"97": "NON_PARTICIPATING",
|
| 258 |
"98": "NON_PARTICIPATING",
|
|
|
|
| 264 |
"attention_bias": false,
|
| 265 |
"attention_dropout": 0.0,
|
| 266 |
"block_list": [
|
| 267 |
+
5812497,
|
| 268 |
+
5812527
|
| 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": 5812444,
|
| 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:f36f0945f1afd697d3dae0f752385ec140202f476002a92fe65a41893f87a226
|
| 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:a2470925e717d7e439db4ea4a0a9a97c25d80f9c0ba45a67d272da01a548f67a
|
| 3 |
size 4400216536
|