Run 5. Outer Step 132. Inner Step 1.
Browse files- config.json +51 -51
- inner_optimizer.pt +1 -1
- model.safetensors +1 -1
config.json
CHANGED
|
@@ -12,16 +12,16 @@
|
|
| 12 |
"105": "NON_PARTICIPATING",
|
| 13 |
"106": "NON_PARTICIPATING",
|
| 14 |
"107": "NOT_ALIVE",
|
| 15 |
-
"108": "
|
| 16 |
"109": "NON_PARTICIPATING",
|
| 17 |
"11": "NOT_ALIVE",
|
| 18 |
"110": "NON_PARTICIPATING",
|
| 19 |
"111": "NON_PARTICIPATING",
|
| 20 |
-
"112": "
|
| 21 |
-
"113": "
|
| 22 |
"114": "NON_PARTICIPATING",
|
| 23 |
"115": "NON_PARTICIPATING",
|
| 24 |
-
"116": "
|
| 25 |
"117": "SUCCESS",
|
| 26 |
"118": "NOT_ALIVE",
|
| 27 |
"119": "NON_PARTICIPATING",
|
|
@@ -31,7 +31,7 @@
|
|
| 31 |
"122": "NON_PARTICIPATING",
|
| 32 |
"123": "NON_PARTICIPATING",
|
| 33 |
"124": "SUCCESS",
|
| 34 |
-
"125": "
|
| 35 |
"126": "SUCCESS",
|
| 36 |
"127": "NOT_ALIVE",
|
| 37 |
"128": "NOT_ALIVE",
|
|
@@ -46,19 +46,19 @@
|
|
| 46 |
"136": "NON_PARTICIPATING",
|
| 47 |
"137": "NON_PARTICIPATING",
|
| 48 |
"138": "NON_PARTICIPATING",
|
| 49 |
-
"139": "
|
| 50 |
-
"14": "
|
| 51 |
-
"140": "
|
| 52 |
"141": "NON_PARTICIPATING",
|
| 53 |
"142": "NON_PARTICIPATING",
|
| 54 |
"143": "NON_PARTICIPATING",
|
| 55 |
"144": "NOT_ALIVE",
|
| 56 |
-
"145": "
|
| 57 |
"146": "NOT_ALIVE",
|
| 58 |
"147": "NON_PARTICIPATING",
|
| 59 |
"148": "SUCCESS",
|
| 60 |
-
"149": "
|
| 61 |
-
"15": "
|
| 62 |
"150": "NOT_ALIVE",
|
| 63 |
"151": "NOT_ALIVE",
|
| 64 |
"152": "NOT_ALIVE",
|
|
@@ -69,16 +69,16 @@
|
|
| 69 |
"157": "NOT_ALIVE",
|
| 70 |
"158": "NOT_ALIVE",
|
| 71 |
"159": "NOT_ALIVE",
|
| 72 |
-
"16": "
|
| 73 |
"160": "NON_PARTICIPATING",
|
| 74 |
"161": "NOT_ALIVE",
|
| 75 |
"162": "NON_PARTICIPATING",
|
| 76 |
"163": "NOT_ALIVE",
|
| 77 |
-
"164": "
|
| 78 |
-
"165": "
|
| 79 |
-
"166": "
|
| 80 |
"167": "NON_PARTICIPATING",
|
| 81 |
-
"168": "
|
| 82 |
"169": "NOT_ALIVE",
|
| 83 |
"17": "NOT_ALIVE",
|
| 84 |
"170": "NOT_ALIVE",
|
|
@@ -95,7 +95,7 @@
|
|
| 95 |
"180": "NOT_ALIVE",
|
| 96 |
"181": "NOT_ALIVE",
|
| 97 |
"182": "NOT_ALIVE",
|
| 98 |
-
"183": "
|
| 99 |
"184": "NON_PARTICIPATING",
|
| 100 |
"185": "NOT_ALIVE",
|
| 101 |
"186": "NOT_ALIVE",
|
|
@@ -104,36 +104,36 @@
|
|
| 104 |
"189": "NOT_ALIVE",
|
| 105 |
"19": "NOT_ALIVE",
|
| 106 |
"190": "NON_PARTICIPATING",
|
| 107 |
-
"191": "
|
| 108 |
"192": "NON_PARTICIPATING",
|
| 109 |
"193": "NOT_ALIVE",
|
| 110 |
-
"194": "
|
| 111 |
"195": "NON_PARTICIPATING",
|
| 112 |
"196": "NON_PARTICIPATING",
|
| 113 |
"197": "SUCCESS",
|
| 114 |
"198": "NON_PARTICIPATING",
|
| 115 |
-
"199": "
|
| 116 |
"2": "NON_PARTICIPATING",
|
| 117 |
"20": "NOT_ALIVE",
|
| 118 |
"200": "NOT_ALIVE",
|
| 119 |
-
"201": "
|
| 120 |
"202": "NOT_ALIVE",
|
| 121 |
"203": "NOT_ALIVE",
|
| 122 |
"204": "SUCCESS",
|
| 123 |
"205": "NON_PARTICIPATING",
|
| 124 |
"206": "SUCCESS",
|
| 125 |
"207": "NOT_ALIVE",
|
| 126 |
-
"208": "
|
| 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": "NOT_ALIVE",
|
| 136 |
-
"217": "
|
| 137 |
"218": "NOT_ALIVE",
|
| 138 |
"219": "NOT_ALIVE",
|
| 139 |
"22": "NOT_ALIVE",
|
|
@@ -146,7 +146,7 @@
|
|
| 146 |
"226": "NOT_ALIVE",
|
| 147 |
"227": "NOT_ALIVE",
|
| 148 |
"228": "NON_PARTICIPATING",
|
| 149 |
-
"229": "
|
| 150 |
"23": "NON_PARTICIPATING",
|
| 151 |
"230": "NON_PARTICIPATING",
|
| 152 |
"231": "NOT_ALIVE",
|
|
@@ -170,27 +170,27 @@
|
|
| 170 |
"248": "NON_PARTICIPATING",
|
| 171 |
"249": "NON_PARTICIPATING",
|
| 172 |
"25": "SUCCESS",
|
| 173 |
-
"250": "
|
| 174 |
-
"251": "
|
| 175 |
"252": "NON_PARTICIPATING",
|
| 176 |
"253": "NOT_ALIVE",
|
| 177 |
-
"254": "
|
| 178 |
"255": "NOT_ALIVE",
|
| 179 |
"26": "NON_PARTICIPATING",
|
| 180 |
"27": "NOT_ALIVE",
|
| 181 |
"28": "NOT_ALIVE",
|
| 182 |
-
"29": "
|
| 183 |
-
"3": "
|
| 184 |
"30": "NON_PARTICIPATING",
|
| 185 |
"31": "NON_PARTICIPATING",
|
| 186 |
-
"32": "
|
| 187 |
"33": "NOT_ALIVE",
|
| 188 |
"34": "NOT_ALIVE",
|
| 189 |
"35": "NON_PARTICIPATING",
|
| 190 |
"36": "NON_PARTICIPATING",
|
| 191 |
"37": "NOT_ALIVE",
|
| 192 |
-
"38": "
|
| 193 |
-
"39": "
|
| 194 |
"4": "NON_PARTICIPATING",
|
| 195 |
"40": "NON_PARTICIPATING",
|
| 196 |
"41": "NON_PARTICIPATING",
|
|
@@ -199,13 +199,13 @@
|
|
| 199 |
"44": "NON_PARTICIPATING",
|
| 200 |
"45": "NOT_ALIVE",
|
| 201 |
"46": "NOT_ALIVE",
|
| 202 |
-
"47": "
|
| 203 |
"48": "NON_PARTICIPATING",
|
| 204 |
"49": "NON_PARTICIPATING",
|
| 205 |
-
"5": "
|
| 206 |
"50": "NOT_ALIVE",
|
| 207 |
"51": "NOT_ALIVE",
|
| 208 |
-
"52": "
|
| 209 |
"53": "NON_PARTICIPATING",
|
| 210 |
"54": "NOT_ALIVE",
|
| 211 |
"55": "NOT_ALIVE",
|
|
@@ -228,33 +228,33 @@
|
|
| 228 |
"70": "NOT_ALIVE",
|
| 229 |
"71": "NON_PARTICIPATING",
|
| 230 |
"72": "NON_PARTICIPATING",
|
| 231 |
-
"73": "
|
| 232 |
"74": "NOT_ALIVE",
|
| 233 |
-
"75": "
|
| 234 |
"76": "NON_PARTICIPATING",
|
| 235 |
-
"77": "
|
| 236 |
"78": "NOT_ALIVE",
|
| 237 |
"79": "NOT_ALIVE",
|
| 238 |
"8": "NOT_ALIVE",
|
| 239 |
-
"80": "
|
| 240 |
-
"81": "
|
| 241 |
"82": "NON_PARTICIPATING",
|
| 242 |
"83": "NOT_ALIVE",
|
| 243 |
"84": "NON_PARTICIPATING",
|
| 244 |
"85": "NON_PARTICIPATING",
|
| 245 |
"86": "NON_PARTICIPATING",
|
| 246 |
-
"87": "
|
| 247 |
-
"88": "
|
| 248 |
"89": "NOT_ALIVE",
|
| 249 |
"9": "NON_PARTICIPATING",
|
| 250 |
"90": "NOT_ALIVE",
|
| 251 |
"91": "NOT_ALIVE",
|
| 252 |
-
"92": "
|
| 253 |
"93": "NOT_ALIVE",
|
| 254 |
-
"94": "
|
| 255 |
"95": "NOT_ALIVE",
|
| 256 |
"96": "NON_PARTICIPATING",
|
| 257 |
-
"97": "
|
| 258 |
"98": "NOT_ALIVE",
|
| 259 |
"99": "NOT_ALIVE"
|
| 260 |
},
|
|
@@ -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",
|
|
|
|
| 12 |
"105": "NON_PARTICIPATING",
|
| 13 |
"106": "NON_PARTICIPATING",
|
| 14 |
"107": "NOT_ALIVE",
|
| 15 |
+
"108": "SUCCESS",
|
| 16 |
"109": "NON_PARTICIPATING",
|
| 17 |
"11": "NOT_ALIVE",
|
| 18 |
"110": "NON_PARTICIPATING",
|
| 19 |
"111": "NON_PARTICIPATING",
|
| 20 |
+
"112": "NON_PARTICIPATING",
|
| 21 |
+
"113": "SUCCESS",
|
| 22 |
"114": "NON_PARTICIPATING",
|
| 23 |
"115": "NON_PARTICIPATING",
|
| 24 |
+
"116": "NOT_ALIVE",
|
| 25 |
"117": "SUCCESS",
|
| 26 |
"118": "NOT_ALIVE",
|
| 27 |
"119": "NON_PARTICIPATING",
|
|
|
|
| 31 |
"122": "NON_PARTICIPATING",
|
| 32 |
"123": "NON_PARTICIPATING",
|
| 33 |
"124": "SUCCESS",
|
| 34 |
+
"125": "NOT_ALIVE",
|
| 35 |
"126": "SUCCESS",
|
| 36 |
"127": "NOT_ALIVE",
|
| 37 |
"128": "NOT_ALIVE",
|
|
|
|
| 46 |
"136": "NON_PARTICIPATING",
|
| 47 |
"137": "NON_PARTICIPATING",
|
| 48 |
"138": "NON_PARTICIPATING",
|
| 49 |
+
"139": "NOT_ALIVE",
|
| 50 |
+
"14": "SUCCESS",
|
| 51 |
+
"140": "NOT_ALIVE",
|
| 52 |
"141": "NON_PARTICIPATING",
|
| 53 |
"142": "NON_PARTICIPATING",
|
| 54 |
"143": "NON_PARTICIPATING",
|
| 55 |
"144": "NOT_ALIVE",
|
| 56 |
+
"145": "NON_PARTICIPATING",
|
| 57 |
"146": "NOT_ALIVE",
|
| 58 |
"147": "NON_PARTICIPATING",
|
| 59 |
"148": "SUCCESS",
|
| 60 |
+
"149": "NOT_ALIVE",
|
| 61 |
+
"15": "NOT_ALIVE",
|
| 62 |
"150": "NOT_ALIVE",
|
| 63 |
"151": "NOT_ALIVE",
|
| 64 |
"152": "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": "NON_PARTICIPATING",
|
| 76 |
"163": "NOT_ALIVE",
|
| 77 |
+
"164": "SUCCESS",
|
| 78 |
+
"165": "SUCCESS",
|
| 79 |
+
"166": "NOT_ALIVE",
|
| 80 |
"167": "NON_PARTICIPATING",
|
| 81 |
+
"168": "NON_PARTICIPATING",
|
| 82 |
"169": "NOT_ALIVE",
|
| 83 |
"17": "NOT_ALIVE",
|
| 84 |
"170": "NOT_ALIVE",
|
|
|
|
| 95 |
"180": "NOT_ALIVE",
|
| 96 |
"181": "NOT_ALIVE",
|
| 97 |
"182": "NOT_ALIVE",
|
| 98 |
+
"183": "NON_PARTICIPATING",
|
| 99 |
"184": "NON_PARTICIPATING",
|
| 100 |
"185": "NOT_ALIVE",
|
| 101 |
"186": "NOT_ALIVE",
|
|
|
|
| 104 |
"189": "NOT_ALIVE",
|
| 105 |
"19": "NOT_ALIVE",
|
| 106 |
"190": "NON_PARTICIPATING",
|
| 107 |
+
"191": "NOT_ALIVE",
|
| 108 |
"192": "NON_PARTICIPATING",
|
| 109 |
"193": "NOT_ALIVE",
|
| 110 |
+
"194": "NON_PARTICIPATING",
|
| 111 |
"195": "NON_PARTICIPATING",
|
| 112 |
"196": "NON_PARTICIPATING",
|
| 113 |
"197": "SUCCESS",
|
| 114 |
"198": "NON_PARTICIPATING",
|
| 115 |
+
"199": "NOT_ALIVE",
|
| 116 |
"2": "NON_PARTICIPATING",
|
| 117 |
"20": "NOT_ALIVE",
|
| 118 |
"200": "NOT_ALIVE",
|
| 119 |
+
"201": "NOT_ALIVE",
|
| 120 |
"202": "NOT_ALIVE",
|
| 121 |
"203": "NOT_ALIVE",
|
| 122 |
"204": "SUCCESS",
|
| 123 |
"205": "NON_PARTICIPATING",
|
| 124 |
"206": "SUCCESS",
|
| 125 |
"207": "NOT_ALIVE",
|
| 126 |
+
"208": "NOT_ALIVE",
|
| 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": "SUCCESS",
|
| 135 |
"216": "NOT_ALIVE",
|
| 136 |
+
"217": "NON_PARTICIPATING",
|
| 137 |
"218": "NOT_ALIVE",
|
| 138 |
"219": "NOT_ALIVE",
|
| 139 |
"22": "NOT_ALIVE",
|
|
|
|
| 146 |
"226": "NOT_ALIVE",
|
| 147 |
"227": "NOT_ALIVE",
|
| 148 |
"228": "NON_PARTICIPATING",
|
| 149 |
+
"229": "NOT_ALIVE",
|
| 150 |
"23": "NON_PARTICIPATING",
|
| 151 |
"230": "NON_PARTICIPATING",
|
| 152 |
"231": "NOT_ALIVE",
|
|
|
|
| 170 |
"248": "NON_PARTICIPATING",
|
| 171 |
"249": "NON_PARTICIPATING",
|
| 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": "NOT_ALIVE",
|
| 179 |
"26": "NON_PARTICIPATING",
|
| 180 |
"27": "NOT_ALIVE",
|
| 181 |
"28": "NOT_ALIVE",
|
| 182 |
+
"29": "NON_PARTICIPATING",
|
| 183 |
+
"3": "NOT_ALIVE",
|
| 184 |
"30": "NON_PARTICIPATING",
|
| 185 |
"31": "NON_PARTICIPATING",
|
| 186 |
+
"32": "SUCCESS",
|
| 187 |
"33": "NOT_ALIVE",
|
| 188 |
"34": "NOT_ALIVE",
|
| 189 |
"35": "NON_PARTICIPATING",
|
| 190 |
"36": "NON_PARTICIPATING",
|
| 191 |
"37": "NOT_ALIVE",
|
| 192 |
+
"38": "NON_PARTICIPATING",
|
| 193 |
+
"39": "SUCCESS",
|
| 194 |
"4": "NON_PARTICIPATING",
|
| 195 |
"40": "NON_PARTICIPATING",
|
| 196 |
"41": "NON_PARTICIPATING",
|
|
|
|
| 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": "NOT_ALIVE",
|
| 208 |
+
"52": "NOT_ALIVE",
|
| 209 |
"53": "NON_PARTICIPATING",
|
| 210 |
"54": "NOT_ALIVE",
|
| 211 |
"55": "NOT_ALIVE",
|
|
|
|
| 228 |
"70": "NOT_ALIVE",
|
| 229 |
"71": "NON_PARTICIPATING",
|
| 230 |
"72": "NON_PARTICIPATING",
|
| 231 |
+
"73": "NON_PARTICIPATING",
|
| 232 |
"74": "NOT_ALIVE",
|
| 233 |
+
"75": "NON_PARTICIPATING",
|
| 234 |
"76": "NON_PARTICIPATING",
|
| 235 |
+
"77": "NON_PARTICIPATING",
|
| 236 |
"78": "NOT_ALIVE",
|
| 237 |
"79": "NOT_ALIVE",
|
| 238 |
"8": "NOT_ALIVE",
|
| 239 |
+
"80": "NON_PARTICIPATING",
|
| 240 |
+
"81": "NON_PARTICIPATING",
|
| 241 |
"82": "NON_PARTICIPATING",
|
| 242 |
"83": "NOT_ALIVE",
|
| 243 |
"84": "NON_PARTICIPATING",
|
| 244 |
"85": "NON_PARTICIPATING",
|
| 245 |
"86": "NON_PARTICIPATING",
|
| 246 |
+
"87": "NON_PARTICIPATING",
|
| 247 |
+
"88": "NOT_ALIVE",
|
| 248 |
"89": "NOT_ALIVE",
|
| 249 |
"9": "NON_PARTICIPATING",
|
| 250 |
"90": "NOT_ALIVE",
|
| 251 |
"91": "NOT_ALIVE",
|
| 252 |
+
"92": "NON_PARTICIPATING",
|
| 253 |
"93": "NOT_ALIVE",
|
| 254 |
+
"94": "SUCCESS",
|
| 255 |
"95": "NOT_ALIVE",
|
| 256 |
"96": "NON_PARTICIPATING",
|
| 257 |
+
"97": "NOT_ALIVE",
|
| 258 |
"98": "NOT_ALIVE",
|
| 259 |
"99": "NOT_ALIVE"
|
| 260 |
},
|
|
|
|
| 264 |
"attention_bias": false,
|
| 265 |
"attention_dropout": 0.0,
|
| 266 |
"block_list": [
|
| 267 |
+
6000863,
|
| 268 |
+
6000882
|
| 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": 1,
|
| 276 |
"intermediate_size": 5632,
|
| 277 |
+
"last_allreduce_block": 6000755,
|
| 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:bfca4ae00c4bb88bfb1fac160066b9d7e1eb62399aef320e5806ae9fceed0b48
|
| 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:eb435d072241b96f7afabb52cdf3b417492939a757df13940f074fea3b899f16
|
| 3 |
size 4400216536
|