Run 5. Outer Step 128. Inner Step 1.
Browse files- config.json +56 -56
- 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": "NOT_ALIVE",
|
| 17 |
"11": "NOT_ALIVE",
|
| 18 |
"110": "NOT_ALIVE",
|
| 19 |
"111": "NON_PARTICIPATING",
|
| 20 |
-
"112": "
|
| 21 |
-
"113": "
|
| 22 |
-
"114": "
|
| 23 |
-
"115": "
|
| 24 |
-
"116": "
|
| 25 |
"117": "NON_PARTICIPATING",
|
| 26 |
"118": "NOT_ALIVE",
|
| 27 |
"119": "NON_PARTICIPATING",
|
|
@@ -30,9 +30,9 @@
|
|
| 30 |
"121": "NON_PARTICIPATING",
|
| 31 |
"122": "NOT_ALIVE",
|
| 32 |
"123": "NOT_ALIVE",
|
| 33 |
-
"124": "
|
| 34 |
-
"125": "
|
| 35 |
-
"126": "
|
| 36 |
"127": "NOT_ALIVE",
|
| 37 |
"128": "NOT_ALIVE",
|
| 38 |
"129": "NOT_ALIVE",
|
|
@@ -47,19 +47,19 @@
|
|
| 47 |
"137": "NON_PARTICIPATING",
|
| 48 |
"138": "NON_PARTICIPATING",
|
| 49 |
"139": "NON_PARTICIPATING",
|
| 50 |
-
"14": "
|
| 51 |
"140": "NON_PARTICIPATING",
|
| 52 |
-
"141": "
|
| 53 |
"142": "NON_PARTICIPATING",
|
| 54 |
-
"143": "
|
| 55 |
-
"144": "
|
| 56 |
"145": "NON_PARTICIPATING",
|
| 57 |
"146": "NOT_ALIVE",
|
| 58 |
"147": "NON_PARTICIPATING",
|
| 59 |
-
"148": "
|
| 60 |
"149": "NOT_ALIVE",
|
| 61 |
"15": "NOT_ALIVE",
|
| 62 |
-
"150": "
|
| 63 |
"151": "NOT_ALIVE",
|
| 64 |
"152": "NOT_ALIVE",
|
| 65 |
"153": "NON_PARTICIPATING",
|
|
@@ -72,13 +72,13 @@
|
|
| 72 |
"16": "NON_PARTICIPATING",
|
| 73 |
"160": "NOT_ALIVE",
|
| 74 |
"161": "NON_PARTICIPATING",
|
| 75 |
-
"162": "
|
| 76 |
"163": "NOT_ALIVE",
|
| 77 |
-
"164": "
|
| 78 |
-
"165": "
|
| 79 |
"166": "NOT_ALIVE",
|
| 80 |
-
"167": "
|
| 81 |
-
"168": "
|
| 82 |
"169": "NOT_ALIVE",
|
| 83 |
"17": "NOT_ALIVE",
|
| 84 |
"170": "NON_PARTICIPATING",
|
|
@@ -96,7 +96,7 @@
|
|
| 96 |
"181": "NOT_ALIVE",
|
| 97 |
"182": "NOT_ALIVE",
|
| 98 |
"183": "SUCCESS",
|
| 99 |
-
"184": "
|
| 100 |
"185": "NOT_ALIVE",
|
| 101 |
"186": "NOT_ALIVE",
|
| 102 |
"187": "NOT_ALIVE",
|
|
@@ -110,9 +110,9 @@
|
|
| 110 |
"194": "NON_PARTICIPATING",
|
| 111 |
"195": "NON_PARTICIPATING",
|
| 112 |
"196": "NON_PARTICIPATING",
|
| 113 |
-
"197": "
|
| 114 |
-
"198": "
|
| 115 |
-
"199": "
|
| 116 |
"2": "NON_PARTICIPATING",
|
| 117 |
"20": "NOT_ALIVE",
|
| 118 |
"200": "NOT_ALIVE",
|
|
@@ -120,24 +120,24 @@
|
|
| 120 |
"202": "NOT_ALIVE",
|
| 121 |
"203": "NOT_ALIVE",
|
| 122 |
"204": "SUCCESS",
|
| 123 |
-
"205": "
|
| 124 |
-
"206": "
|
| 125 |
"207": "NOT_ALIVE",
|
| 126 |
"208": "NON_PARTICIPATING",
|
| 127 |
"209": "NOT_ALIVE",
|
| 128 |
"21": "NOT_ALIVE",
|
| 129 |
"210": "NOT_ALIVE",
|
| 130 |
-
"211": "
|
| 131 |
-
"212": "
|
| 132 |
-
"213": "
|
| 133 |
"214": "NOT_ALIVE",
|
| 134 |
-
"215": "
|
| 135 |
"216": "NOT_ALIVE",
|
| 136 |
"217": "NON_PARTICIPATING",
|
| 137 |
-
"218": "
|
| 138 |
"219": "NOT_ALIVE",
|
| 139 |
"22": "NOT_ALIVE",
|
| 140 |
-
"220": "
|
| 141 |
"221": "NOT_ALIVE",
|
| 142 |
"222": "NOT_ALIVE",
|
| 143 |
"223": "NOT_ALIVE",
|
|
@@ -160,26 +160,26 @@
|
|
| 160 |
"239": "NOT_ALIVE",
|
| 161 |
"24": "NOT_ALIVE",
|
| 162 |
"240": "NOT_ALIVE",
|
| 163 |
-
"241": "
|
| 164 |
"242": "NOT_ALIVE",
|
| 165 |
-
"243": "
|
| 166 |
"244": "NOT_ALIVE",
|
| 167 |
"245": "NON_PARTICIPATING",
|
| 168 |
"246": "NON_PARTICIPATING",
|
| 169 |
-
"247": "
|
| 170 |
"248": "NOT_ALIVE",
|
| 171 |
"249": "NOT_ALIVE",
|
| 172 |
-
"25": "
|
| 173 |
-
"250": "
|
| 174 |
"251": "NON_PARTICIPATING",
|
| 175 |
-
"252": "
|
| 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": "NON_PARTICIPATING",
|
| 184 |
"30": "SUCCESS",
|
| 185 |
"31": "NOT_ALIVE",
|
|
@@ -189,10 +189,10 @@
|
|
| 189 |
"35": "NON_PARTICIPATING",
|
| 190 |
"36": "NOT_ALIVE",
|
| 191 |
"37": "NOT_ALIVE",
|
| 192 |
-
"38": "
|
| 193 |
-
"39": "
|
| 194 |
"4": "NON_PARTICIPATING",
|
| 195 |
-
"40": "
|
| 196 |
"41": "NOT_ALIVE",
|
| 197 |
"42": "NOT_ALIVE",
|
| 198 |
"43": "NOT_ALIVE",
|
|
@@ -201,8 +201,8 @@
|
|
| 201 |
"46": "NOT_ALIVE",
|
| 202 |
"47": "SUCCESS",
|
| 203 |
"48": "NON_PARTICIPATING",
|
| 204 |
-
"49": "
|
| 205 |
-
"5": "
|
| 206 |
"50": "NOT_ALIVE",
|
| 207 |
"51": "NOT_ALIVE",
|
| 208 |
"52": "SUCCESS",
|
|
@@ -212,9 +212,9 @@
|
|
| 212 |
"56": "NOT_ALIVE",
|
| 213 |
"57": "NON_PARTICIPATING",
|
| 214 |
"58": "NON_PARTICIPATING",
|
| 215 |
-
"59": "
|
| 216 |
"6": "NON_PARTICIPATING",
|
| 217 |
-
"60": "
|
| 218 |
"61": "NON_PARTICIPATING",
|
| 219 |
"62": "NON_PARTICIPATING",
|
| 220 |
"63": "NON_PARTICIPATING",
|
|
@@ -231,8 +231,8 @@
|
|
| 231 |
"73": "NOT_ALIVE",
|
| 232 |
"74": "NON_PARTICIPATING",
|
| 233 |
"75": "NON_PARTICIPATING",
|
| 234 |
-
"76": "
|
| 235 |
-
"77": "
|
| 236 |
"78": "NOT_ALIVE",
|
| 237 |
"79": "NOT_ALIVE",
|
| 238 |
"8": "NOT_ALIVE",
|
|
@@ -251,11 +251,11 @@
|
|
| 251 |
"91": "NOT_ALIVE",
|
| 252 |
"92": "SUCCESS",
|
| 253 |
"93": "NOT_ALIVE",
|
| 254 |
-
"94": "
|
| 255 |
"95": "NOT_ALIVE",
|
| 256 |
"96": "NON_PARTICIPATING",
|
| 257 |
-
"97": "
|
| 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",
|
|
|
|
| 12 |
"105": "NON_PARTICIPATING",
|
| 13 |
"106": "NON_PARTICIPATING",
|
| 14 |
"107": "NOT_ALIVE",
|
| 15 |
+
"108": "NON_PARTICIPATING",
|
| 16 |
"109": "NOT_ALIVE",
|
| 17 |
"11": "NOT_ALIVE",
|
| 18 |
"110": "NOT_ALIVE",
|
| 19 |
"111": "NON_PARTICIPATING",
|
| 20 |
+
"112": "NON_PARTICIPATING",
|
| 21 |
+
"113": "SUCCESS",
|
| 22 |
+
"114": "NOT_ALIVE",
|
| 23 |
+
"115": "NOT_ALIVE",
|
| 24 |
+
"116": "NOT_ALIVE",
|
| 25 |
"117": "NON_PARTICIPATING",
|
| 26 |
"118": "NOT_ALIVE",
|
| 27 |
"119": "NON_PARTICIPATING",
|
|
|
|
| 30 |
"121": "NON_PARTICIPATING",
|
| 31 |
"122": "NOT_ALIVE",
|
| 32 |
"123": "NOT_ALIVE",
|
| 33 |
+
"124": "SUCCESS",
|
| 34 |
+
"125": "NOT_ALIVE",
|
| 35 |
+
"126": "SUCCESS",
|
| 36 |
"127": "NOT_ALIVE",
|
| 37 |
"128": "NOT_ALIVE",
|
| 38 |
"129": "NOT_ALIVE",
|
|
|
|
| 47 |
"137": "NON_PARTICIPATING",
|
| 48 |
"138": "NON_PARTICIPATING",
|
| 49 |
"139": "NON_PARTICIPATING",
|
| 50 |
+
"14": "SUCCESS",
|
| 51 |
"140": "NON_PARTICIPATING",
|
| 52 |
+
"141": "NON_PARTICIPATING",
|
| 53 |
"142": "NON_PARTICIPATING",
|
| 54 |
+
"143": "NOT_ALIVE",
|
| 55 |
+
"144": "NON_PARTICIPATING",
|
| 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": "NON_PARTICIPATING",
|
| 63 |
"151": "NOT_ALIVE",
|
| 64 |
"152": "NOT_ALIVE",
|
| 65 |
"153": "NON_PARTICIPATING",
|
|
|
|
| 72 |
"16": "NON_PARTICIPATING",
|
| 73 |
"160": "NOT_ALIVE",
|
| 74 |
"161": "NON_PARTICIPATING",
|
| 75 |
+
"162": "NOT_ALIVE",
|
| 76 |
"163": "NOT_ALIVE",
|
| 77 |
+
"164": "NON_PARTICIPATING",
|
| 78 |
+
"165": "NOT_ALIVE",
|
| 79 |
"166": "NOT_ALIVE",
|
| 80 |
+
"167": "NON_PARTICIPATING",
|
| 81 |
+
"168": "SUCCESS",
|
| 82 |
"169": "NOT_ALIVE",
|
| 83 |
"17": "NOT_ALIVE",
|
| 84 |
"170": "NON_PARTICIPATING",
|
|
|
|
| 96 |
"181": "NOT_ALIVE",
|
| 97 |
"182": "NOT_ALIVE",
|
| 98 |
"183": "SUCCESS",
|
| 99 |
+
"184": "NOT_ALIVE",
|
| 100 |
"185": "NOT_ALIVE",
|
| 101 |
"186": "NOT_ALIVE",
|
| 102 |
"187": "NOT_ALIVE",
|
|
|
|
| 110 |
"194": "NON_PARTICIPATING",
|
| 111 |
"195": "NON_PARTICIPATING",
|
| 112 |
"196": "NON_PARTICIPATING",
|
| 113 |
+
"197": "NON_PARTICIPATING",
|
| 114 |
+
"198": "NON_PARTICIPATING",
|
| 115 |
+
"199": "NON_PARTICIPATING",
|
| 116 |
"2": "NON_PARTICIPATING",
|
| 117 |
"20": "NOT_ALIVE",
|
| 118 |
"200": "NOT_ALIVE",
|
|
|
|
| 120 |
"202": "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": "NOT_ALIVE",
|
| 130 |
+
"211": "NON_PARTICIPATING",
|
| 131 |
+
"212": "SUCCESS",
|
| 132 |
+
"213": "NON_PARTICIPATING",
|
| 133 |
"214": "NOT_ALIVE",
|
| 134 |
+
"215": "NOT_ALIVE",
|
| 135 |
"216": "NOT_ALIVE",
|
| 136 |
"217": "NON_PARTICIPATING",
|
| 137 |
+
"218": "NON_PARTICIPATING",
|
| 138 |
"219": "NOT_ALIVE",
|
| 139 |
"22": "NOT_ALIVE",
|
| 140 |
+
"220": "NOT_ALIVE",
|
| 141 |
"221": "NOT_ALIVE",
|
| 142 |
"222": "NOT_ALIVE",
|
| 143 |
"223": "NOT_ALIVE",
|
|
|
|
| 160 |
"239": "NOT_ALIVE",
|
| 161 |
"24": "NOT_ALIVE",
|
| 162 |
"240": "NOT_ALIVE",
|
| 163 |
+
"241": "NON_PARTICIPATING",
|
| 164 |
"242": "NOT_ALIVE",
|
| 165 |
+
"243": "NOT_ALIVE",
|
| 166 |
"244": "NOT_ALIVE",
|
| 167 |
"245": "NON_PARTICIPATING",
|
| 168 |
"246": "NON_PARTICIPATING",
|
| 169 |
+
"247": "NOT_ALIVE",
|
| 170 |
"248": "NOT_ALIVE",
|
| 171 |
"249": "NOT_ALIVE",
|
| 172 |
+
"25": "NON_PARTICIPATING",
|
| 173 |
+
"250": "SUCCESS",
|
| 174 |
"251": "NON_PARTICIPATING",
|
| 175 |
+
"252": "NON_PARTICIPATING",
|
| 176 |
"253": "NOT_ALIVE",
|
| 177 |
+
"254": "SUCCESS",
|
| 178 |
"255": "NOT_ALIVE",
|
| 179 |
"26": "NON_PARTICIPATING",
|
| 180 |
"27": "NOT_ALIVE",
|
| 181 |
"28": "NOT_ALIVE",
|
| 182 |
+
"29": "NON_PARTICIPATING",
|
| 183 |
"3": "NON_PARTICIPATING",
|
| 184 |
"30": "SUCCESS",
|
| 185 |
"31": "NOT_ALIVE",
|
|
|
|
| 189 |
"35": "NON_PARTICIPATING",
|
| 190 |
"36": "NOT_ALIVE",
|
| 191 |
"37": "NOT_ALIVE",
|
| 192 |
+
"38": "NOT_ALIVE",
|
| 193 |
+
"39": "NOT_ALIVE",
|
| 194 |
"4": "NON_PARTICIPATING",
|
| 195 |
+
"40": "FAIL",
|
| 196 |
"41": "NOT_ALIVE",
|
| 197 |
"42": "NOT_ALIVE",
|
| 198 |
"43": "NOT_ALIVE",
|
|
|
|
| 201 |
"46": "NOT_ALIVE",
|
| 202 |
"47": "SUCCESS",
|
| 203 |
"48": "NON_PARTICIPATING",
|
| 204 |
+
"49": "NON_PARTICIPATING",
|
| 205 |
+
"5": "SUCCESS",
|
| 206 |
"50": "NOT_ALIVE",
|
| 207 |
"51": "NOT_ALIVE",
|
| 208 |
"52": "SUCCESS",
|
|
|
|
| 212 |
"56": "NOT_ALIVE",
|
| 213 |
"57": "NON_PARTICIPATING",
|
| 214 |
"58": "NON_PARTICIPATING",
|
| 215 |
+
"59": "NOT_ALIVE",
|
| 216 |
"6": "NON_PARTICIPATING",
|
| 217 |
+
"60": "NOT_ALIVE",
|
| 218 |
"61": "NON_PARTICIPATING",
|
| 219 |
"62": "NON_PARTICIPATING",
|
| 220 |
"63": "NON_PARTICIPATING",
|
|
|
|
| 231 |
"73": "NOT_ALIVE",
|
| 232 |
"74": "NON_PARTICIPATING",
|
| 233 |
"75": "NON_PARTICIPATING",
|
| 234 |
+
"76": "NOT_ALIVE",
|
| 235 |
+
"77": "SUCCESS",
|
| 236 |
"78": "NOT_ALIVE",
|
| 237 |
"79": "NOT_ALIVE",
|
| 238 |
"8": "NOT_ALIVE",
|
|
|
|
| 251 |
"91": "NOT_ALIVE",
|
| 252 |
"92": "SUCCESS",
|
| 253 |
"93": "NOT_ALIVE",
|
| 254 |
+
"94": "NON_PARTICIPATING",
|
| 255 |
"95": "NOT_ALIVE",
|
| 256 |
"96": "NON_PARTICIPATING",
|
| 257 |
+
"97": "NOT_ALIVE",
|
| 258 |
+
"98": "NOT_ALIVE",
|
| 259 |
"99": "NOT_ALIVE"
|
| 260 |
},
|
| 261 |
"architectures": [
|
|
|
|
| 264 |
"attention_bias": false,
|
| 265 |
"attention_dropout": 0.0,
|
| 266 |
"block_list": [
|
| 267 |
+
5994469,
|
| 268 |
+
5994503
|
| 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": 5994342,
|
| 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:b1d2e0734effff86cd7cc6b41ba9613afc6db551fe7215bdcd1157e9bb5b826d
|
| 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 sha256:649a52659c4a715abda0541295a871b9af1adc0824a77316b0b2f5da4cc23e59
|
| 3 |
size 4400216536
|