Run 5. Outer Step 58. Inner Step 2.
Browse files- config.json +50 -50
- inner_optimizer.pt +1 -1
- model.safetensors +1 -1
config.json
CHANGED
|
@@ -7,7 +7,7 @@
|
|
| 7 |
"100": "NOT_ALIVE",
|
| 8 |
"101": "NON_PARTICIPATING",
|
| 9 |
"102": "NOT_ALIVE",
|
| 10 |
-
"103": "
|
| 11 |
"104": "NOT_ALIVE",
|
| 12 |
"105": "SUCCESS",
|
| 13 |
"106": "NOT_ALIVE",
|
|
@@ -20,12 +20,12 @@
|
|
| 20 |
"112": "NOT_ALIVE",
|
| 21 |
"113": "NON_PARTICIPATING",
|
| 22 |
"114": "NON_PARTICIPATING",
|
| 23 |
-
"115": "
|
| 24 |
-
"116": "
|
| 25 |
"117": "NOT_ALIVE",
|
| 26 |
-
"118": "
|
| 27 |
-
"119": "
|
| 28 |
-
"12": "
|
| 29 |
"120": "NOT_ALIVE",
|
| 30 |
"121": "NOT_ALIVE",
|
| 31 |
"122": "SUCCESS",
|
|
@@ -33,7 +33,7 @@
|
|
| 33 |
"124": "SUCCESS",
|
| 34 |
"125": "NOT_ALIVE",
|
| 35 |
"126": "NOT_ALIVE",
|
| 36 |
-
"127": "
|
| 37 |
"128": "NOT_ALIVE",
|
| 38 |
"129": "SUCCESS",
|
| 39 |
"13": "NOT_ALIVE",
|
|
@@ -51,21 +51,21 @@
|
|
| 51 |
"140": "NOT_ALIVE",
|
| 52 |
"141": "SUCCESS",
|
| 53 |
"142": "SUCCESS",
|
| 54 |
-
"143": "
|
| 55 |
-
"144": "
|
| 56 |
"145": "NOT_ALIVE",
|
| 57 |
-
"146": "
|
| 58 |
-
"147": "
|
| 59 |
-
"148": "
|
| 60 |
-
"149": "
|
| 61 |
"15": "NON_PARTICIPATING",
|
| 62 |
"150": "SUCCESS",
|
| 63 |
"151": "NOT_ALIVE",
|
| 64 |
-
"152": "
|
| 65 |
-
"153": "
|
| 66 |
-
"154": "
|
| 67 |
"155": "SUCCESS",
|
| 68 |
-
"156": "
|
| 69 |
"157": "NON_PARTICIPATING",
|
| 70 |
"158": "NOT_ALIVE",
|
| 71 |
"159": "NOT_ALIVE",
|
|
@@ -75,8 +75,8 @@
|
|
| 75 |
"162": "NON_PARTICIPATING",
|
| 76 |
"163": "NOT_ALIVE",
|
| 77 |
"164": "SUCCESS",
|
| 78 |
-
"165": "
|
| 79 |
-
"166": "
|
| 80 |
"167": "NON_PARTICIPATING",
|
| 81 |
"168": "SUCCESS",
|
| 82 |
"169": "NOT_ALIVE",
|
|
@@ -92,7 +92,7 @@
|
|
| 92 |
"178": "NOT_ALIVE",
|
| 93 |
"179": "NOT_ALIVE",
|
| 94 |
"18": "SUCCESS",
|
| 95 |
-
"180": "
|
| 96 |
"181": "SUCCESS",
|
| 97 |
"182": "NON_PARTICIPATING",
|
| 98 |
"183": "SUCCESS",
|
|
@@ -112,14 +112,14 @@
|
|
| 112 |
"196": "NOT_ALIVE",
|
| 113 |
"197": "NOT_ALIVE",
|
| 114 |
"198": "NOT_ALIVE",
|
| 115 |
-
"199": "
|
| 116 |
"2": "NOT_ALIVE",
|
| 117 |
"20": "NON_PARTICIPATING",
|
| 118 |
"200": "NON_PARTICIPATING",
|
| 119 |
"201": "NON_PARTICIPATING",
|
| 120 |
-
"202": "
|
| 121 |
"203": "NON_PARTICIPATING",
|
| 122 |
-
"204": "
|
| 123 |
"205": "NOT_ALIVE",
|
| 124 |
"206": "SUCCESS",
|
| 125 |
"207": "NOT_ALIVE",
|
|
@@ -127,37 +127,37 @@
|
|
| 127 |
"209": "SUCCESS",
|
| 128 |
"21": "SUCCESS",
|
| 129 |
"210": "SUCCESS",
|
| 130 |
-
"211": "
|
| 131 |
"212": "NOT_ALIVE",
|
| 132 |
"213": "NON_PARTICIPATING",
|
| 133 |
"214": "NON_PARTICIPATING",
|
| 134 |
-
"215": "
|
| 135 |
"216": "NOT_ALIVE",
|
| 136 |
"217": "NON_PARTICIPATING",
|
| 137 |
"218": "NON_PARTICIPATING",
|
| 138 |
"219": "NOT_ALIVE",
|
| 139 |
"22": "NON_PARTICIPATING",
|
| 140 |
-
"220": "
|
| 141 |
"221": "SUCCESS",
|
| 142 |
-
"222": "
|
| 143 |
"223": "NOT_ALIVE",
|
| 144 |
"224": "NOT_ALIVE",
|
| 145 |
"225": "NOT_ALIVE",
|
| 146 |
"226": "NOT_ALIVE",
|
| 147 |
-
"227": "
|
| 148 |
"228": "NOT_ALIVE",
|
| 149 |
"229": "NOT_ALIVE",
|
| 150 |
"23": "NON_PARTICIPATING",
|
| 151 |
"230": "NOT_ALIVE",
|
| 152 |
-
"231": "
|
| 153 |
"232": "NOT_ALIVE",
|
| 154 |
"233": "NOT_ALIVE",
|
| 155 |
-
"234": "
|
| 156 |
"235": "NOT_ALIVE",
|
| 157 |
-
"236": "
|
| 158 |
"237": "NOT_ALIVE",
|
| 159 |
"238": "NON_PARTICIPATING",
|
| 160 |
-
"239": "
|
| 161 |
"24": "SUCCESS",
|
| 162 |
"240": "NON_PARTICIPATING",
|
| 163 |
"241": "NOT_ALIVE",
|
|
@@ -168,47 +168,47 @@
|
|
| 168 |
"246": "NON_PARTICIPATING",
|
| 169 |
"247": "SUCCESS",
|
| 170 |
"248": "NON_PARTICIPATING",
|
| 171 |
-
"249": "
|
| 172 |
"25": "SUCCESS",
|
| 173 |
"250": "NON_PARTICIPATING",
|
| 174 |
"251": "NOT_ALIVE",
|
| 175 |
-
"252": "
|
| 176 |
"253": "NOT_ALIVE",
|
| 177 |
"254": "NOT_ALIVE",
|
| 178 |
-
"255": "
|
| 179 |
"26": "SUCCESS",
|
| 180 |
"27": "SUCCESS",
|
| 181 |
-
"28": "
|
| 182 |
"29": "NOT_ALIVE",
|
| 183 |
"3": "SUCCESS",
|
| 184 |
"30": "NOT_ALIVE",
|
| 185 |
"31": "NON_PARTICIPATING",
|
| 186 |
"32": "NOT_ALIVE",
|
| 187 |
"33": "NOT_ALIVE",
|
| 188 |
-
"34": "
|
| 189 |
"35": "NOT_ALIVE",
|
| 190 |
"36": "NON_PARTICIPATING",
|
| 191 |
-
"37": "
|
| 192 |
"38": "SUCCESS",
|
| 193 |
"39": "SUCCESS",
|
| 194 |
"4": "NOT_ALIVE",
|
| 195 |
"40": "SUCCESS",
|
| 196 |
-
"41": "
|
| 197 |
"42": "SUCCESS",
|
| 198 |
-
"43": "
|
| 199 |
"44": "NOT_ALIVE",
|
| 200 |
-
"45": "
|
| 201 |
"46": "NOT_ALIVE",
|
| 202 |
-
"47": "
|
| 203 |
"48": "SUCCESS",
|
| 204 |
"49": "NOT_ALIVE",
|
| 205 |
"5": "NOT_ALIVE",
|
| 206 |
-
"50": "
|
| 207 |
"51": "NOT_ALIVE",
|
| 208 |
"52": "NOT_ALIVE",
|
| 209 |
"53": "NON_PARTICIPATING",
|
| 210 |
"54": "SUCCESS",
|
| 211 |
-
"55": "
|
| 212 |
"56": "SUCCESS",
|
| 213 |
"57": "SUCCESS",
|
| 214 |
"58": "SUCCESS",
|
|
@@ -234,7 +234,7 @@
|
|
| 234 |
"76": "SUCCESS",
|
| 235 |
"77": "NOT_ALIVE",
|
| 236 |
"78": "NOT_ALIVE",
|
| 237 |
-
"79": "
|
| 238 |
"8": "NOT_ALIVE",
|
| 239 |
"80": "NON_PARTICIPATING",
|
| 240 |
"81": "NOT_ALIVE",
|
|
@@ -255,7 +255,7 @@
|
|
| 255 |
"95": "NOT_ALIVE",
|
| 256 |
"96": "NON_PARTICIPATING",
|
| 257 |
"97": "SUCCESS",
|
| 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",
|
|
|
|
| 7 |
"100": "NOT_ALIVE",
|
| 8 |
"101": "NON_PARTICIPATING",
|
| 9 |
"102": "NOT_ALIVE",
|
| 10 |
+
"103": "NOT_ALIVE",
|
| 11 |
"104": "NOT_ALIVE",
|
| 12 |
"105": "SUCCESS",
|
| 13 |
"106": "NOT_ALIVE",
|
|
|
|
| 20 |
"112": "NOT_ALIVE",
|
| 21 |
"113": "NON_PARTICIPATING",
|
| 22 |
"114": "NON_PARTICIPATING",
|
| 23 |
+
"115": "NOT_ALIVE",
|
| 24 |
+
"116": "NON_PARTICIPATING",
|
| 25 |
"117": "NOT_ALIVE",
|
| 26 |
+
"118": "NOT_ALIVE",
|
| 27 |
+
"119": "NOT_ALIVE",
|
| 28 |
+
"12": "NOT_ALIVE",
|
| 29 |
"120": "NOT_ALIVE",
|
| 30 |
"121": "NOT_ALIVE",
|
| 31 |
"122": "SUCCESS",
|
|
|
|
| 33 |
"124": "SUCCESS",
|
| 34 |
"125": "NOT_ALIVE",
|
| 35 |
"126": "NOT_ALIVE",
|
| 36 |
+
"127": "NON_PARTICIPATING",
|
| 37 |
"128": "NOT_ALIVE",
|
| 38 |
"129": "SUCCESS",
|
| 39 |
"13": "NOT_ALIVE",
|
|
|
|
| 51 |
"140": "NOT_ALIVE",
|
| 52 |
"141": "SUCCESS",
|
| 53 |
"142": "SUCCESS",
|
| 54 |
+
"143": "NON_PARTICIPATING",
|
| 55 |
+
"144": "NON_PARTICIPATING",
|
| 56 |
"145": "NOT_ALIVE",
|
| 57 |
+
"146": "NOT_ALIVE",
|
| 58 |
+
"147": "NOT_ALIVE",
|
| 59 |
+
"148": "SUCCESS",
|
| 60 |
+
"149": "NOT_ALIVE",
|
| 61 |
"15": "NON_PARTICIPATING",
|
| 62 |
"150": "SUCCESS",
|
| 63 |
"151": "NOT_ALIVE",
|
| 64 |
+
"152": "NON_PARTICIPATING",
|
| 65 |
+
"153": "SUCCESS",
|
| 66 |
+
"154": "NON_PARTICIPATING",
|
| 67 |
"155": "SUCCESS",
|
| 68 |
+
"156": "NON_PARTICIPATING",
|
| 69 |
"157": "NON_PARTICIPATING",
|
| 70 |
"158": "NOT_ALIVE",
|
| 71 |
"159": "NOT_ALIVE",
|
|
|
|
| 75 |
"162": "NON_PARTICIPATING",
|
| 76 |
"163": "NOT_ALIVE",
|
| 77 |
"164": "SUCCESS",
|
| 78 |
+
"165": "SUCCESS",
|
| 79 |
+
"166": "NOT_ALIVE",
|
| 80 |
"167": "NON_PARTICIPATING",
|
| 81 |
"168": "SUCCESS",
|
| 82 |
"169": "NOT_ALIVE",
|
|
|
|
| 92 |
"178": "NOT_ALIVE",
|
| 93 |
"179": "NOT_ALIVE",
|
| 94 |
"18": "SUCCESS",
|
| 95 |
+
"180": "NON_PARTICIPATING",
|
| 96 |
"181": "SUCCESS",
|
| 97 |
"182": "NON_PARTICIPATING",
|
| 98 |
"183": "SUCCESS",
|
|
|
|
| 112 |
"196": "NOT_ALIVE",
|
| 113 |
"197": "NOT_ALIVE",
|
| 114 |
"198": "NOT_ALIVE",
|
| 115 |
+
"199": "NON_PARTICIPATING",
|
| 116 |
"2": "NOT_ALIVE",
|
| 117 |
"20": "NON_PARTICIPATING",
|
| 118 |
"200": "NON_PARTICIPATING",
|
| 119 |
"201": "NON_PARTICIPATING",
|
| 120 |
+
"202": "NON_PARTICIPATING",
|
| 121 |
"203": "NON_PARTICIPATING",
|
| 122 |
+
"204": "NOT_ALIVE",
|
| 123 |
"205": "NOT_ALIVE",
|
| 124 |
"206": "SUCCESS",
|
| 125 |
"207": "NOT_ALIVE",
|
|
|
|
| 127 |
"209": "SUCCESS",
|
| 128 |
"21": "SUCCESS",
|
| 129 |
"210": "SUCCESS",
|
| 130 |
+
"211": "NOT_ALIVE",
|
| 131 |
"212": "NOT_ALIVE",
|
| 132 |
"213": "NON_PARTICIPATING",
|
| 133 |
"214": "NON_PARTICIPATING",
|
| 134 |
+
"215": "NON_PARTICIPATING",
|
| 135 |
"216": "NOT_ALIVE",
|
| 136 |
"217": "NON_PARTICIPATING",
|
| 137 |
"218": "NON_PARTICIPATING",
|
| 138 |
"219": "NOT_ALIVE",
|
| 139 |
"22": "NON_PARTICIPATING",
|
| 140 |
+
"220": "SUCCESS",
|
| 141 |
"221": "SUCCESS",
|
| 142 |
+
"222": "NON_PARTICIPATING",
|
| 143 |
"223": "NOT_ALIVE",
|
| 144 |
"224": "NOT_ALIVE",
|
| 145 |
"225": "NOT_ALIVE",
|
| 146 |
"226": "NOT_ALIVE",
|
| 147 |
+
"227": "NOT_ALIVE",
|
| 148 |
"228": "NOT_ALIVE",
|
| 149 |
"229": "NOT_ALIVE",
|
| 150 |
"23": "NON_PARTICIPATING",
|
| 151 |
"230": "NOT_ALIVE",
|
| 152 |
+
"231": "NON_PARTICIPATING",
|
| 153 |
"232": "NOT_ALIVE",
|
| 154 |
"233": "NOT_ALIVE",
|
| 155 |
+
"234": "NON_PARTICIPATING",
|
| 156 |
"235": "NOT_ALIVE",
|
| 157 |
+
"236": "NON_PARTICIPATING",
|
| 158 |
"237": "NOT_ALIVE",
|
| 159 |
"238": "NON_PARTICIPATING",
|
| 160 |
+
"239": "NON_PARTICIPATING",
|
| 161 |
"24": "SUCCESS",
|
| 162 |
"240": "NON_PARTICIPATING",
|
| 163 |
"241": "NOT_ALIVE",
|
|
|
|
| 168 |
"246": "NON_PARTICIPATING",
|
| 169 |
"247": "SUCCESS",
|
| 170 |
"248": "NON_PARTICIPATING",
|
| 171 |
+
"249": "NOT_ALIVE",
|
| 172 |
"25": "SUCCESS",
|
| 173 |
"250": "NON_PARTICIPATING",
|
| 174 |
"251": "NOT_ALIVE",
|
| 175 |
+
"252": "NON_PARTICIPATING",
|
| 176 |
"253": "NOT_ALIVE",
|
| 177 |
"254": "NOT_ALIVE",
|
| 178 |
+
"255": "NON_PARTICIPATING",
|
| 179 |
"26": "SUCCESS",
|
| 180 |
"27": "SUCCESS",
|
| 181 |
+
"28": "NOT_ALIVE",
|
| 182 |
"29": "NOT_ALIVE",
|
| 183 |
"3": "SUCCESS",
|
| 184 |
"30": "NOT_ALIVE",
|
| 185 |
"31": "NON_PARTICIPATING",
|
| 186 |
"32": "NOT_ALIVE",
|
| 187 |
"33": "NOT_ALIVE",
|
| 188 |
+
"34": "NOT_ALIVE",
|
| 189 |
"35": "NOT_ALIVE",
|
| 190 |
"36": "NON_PARTICIPATING",
|
| 191 |
+
"37": "NON_PARTICIPATING",
|
| 192 |
"38": "SUCCESS",
|
| 193 |
"39": "SUCCESS",
|
| 194 |
"4": "NOT_ALIVE",
|
| 195 |
"40": "SUCCESS",
|
| 196 |
+
"41": "SUCCESS",
|
| 197 |
"42": "SUCCESS",
|
| 198 |
+
"43": "NOT_ALIVE",
|
| 199 |
"44": "NOT_ALIVE",
|
| 200 |
+
"45": "SUCCESS",
|
| 201 |
"46": "NOT_ALIVE",
|
| 202 |
+
"47": "NOT_ALIVE",
|
| 203 |
"48": "SUCCESS",
|
| 204 |
"49": "NOT_ALIVE",
|
| 205 |
"5": "NOT_ALIVE",
|
| 206 |
+
"50": "NON_PARTICIPATING",
|
| 207 |
"51": "NOT_ALIVE",
|
| 208 |
"52": "NOT_ALIVE",
|
| 209 |
"53": "NON_PARTICIPATING",
|
| 210 |
"54": "SUCCESS",
|
| 211 |
+
"55": "SUCCESS",
|
| 212 |
"56": "SUCCESS",
|
| 213 |
"57": "SUCCESS",
|
| 214 |
"58": "SUCCESS",
|
|
|
|
| 234 |
"76": "SUCCESS",
|
| 235 |
"77": "NOT_ALIVE",
|
| 236 |
"78": "NOT_ALIVE",
|
| 237 |
+
"79": "NON_PARTICIPATING",
|
| 238 |
"8": "NOT_ALIVE",
|
| 239 |
"80": "NON_PARTICIPATING",
|
| 240 |
"81": "NOT_ALIVE",
|
|
|
|
| 255 |
"95": "NOT_ALIVE",
|
| 256 |
"96": "NON_PARTICIPATING",
|
| 257 |
"97": "SUCCESS",
|
| 258 |
+
"98": "SUCCESS",
|
| 259 |
"99": "NOT_ALIVE"
|
| 260 |
},
|
| 261 |
"architectures": [
|
|
|
|
| 264 |
"attention_bias": false,
|
| 265 |
"attention_dropout": 0.0,
|
| 266 |
"block_list": [
|
| 267 |
+
5896840,
|
| 268 |
+
5896886
|
| 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": 5896685,
|
| 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 sha256:9e6bde2d5b92cd88dec3d542eabd87f9d88a97203ca2a85abc90adb324c8b872
|
| 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 sha256:0d535667c20bcb6dbc7695757f4f1b38c40a2ce05113436f04511f5980c3401d
|
| 3 |
size 4400216536
|