Run 6. Outer Step 33. Inner Step 5.
Browse files- config.json +35 -35
- gradients.pt +1 -1
- inner_optimizer.pt +1 -1
- model.safetensors +1 -1
config.json
CHANGED
|
@@ -4,9 +4,9 @@
|
|
| 4 |
"0": "NOT_ALIVE",
|
| 5 |
"1": "NOT_ALIVE",
|
| 6 |
"10": "NOT_ALIVE",
|
| 7 |
-
"100": "
|
| 8 |
"101": "NOT_ALIVE",
|
| 9 |
-
"102": "
|
| 10 |
"103": "NON_PARTICIPATING",
|
| 11 |
"104": "NOT_ALIVE",
|
| 12 |
"105": "NON_PARTICIPATING",
|
|
@@ -21,16 +21,16 @@
|
|
| 21 |
"113": "SUCCESS",
|
| 22 |
"114": "NOT_ALIVE",
|
| 23 |
"115": "NON_PARTICIPATING",
|
| 24 |
-
"116": "
|
| 25 |
"117": "NON_PARTICIPATING",
|
| 26 |
"118": "NON_PARTICIPATING",
|
| 27 |
"119": "NOT_ALIVE",
|
| 28 |
"12": "NOT_ALIVE",
|
| 29 |
"120": "NOT_ALIVE",
|
| 30 |
-
"121": "
|
| 31 |
"122": "NON_PARTICIPATING",
|
| 32 |
"123": "NON_PARTICIPATING",
|
| 33 |
-
"124": "
|
| 34 |
"125": "NOT_ALIVE",
|
| 35 |
"126": "SUCCESS",
|
| 36 |
"127": "NOT_ALIVE",
|
|
@@ -49,33 +49,33 @@
|
|
| 49 |
"139": "NOT_ALIVE",
|
| 50 |
"14": "NOT_ALIVE",
|
| 51 |
"140": "SUCCESS",
|
| 52 |
-
"141": "
|
| 53 |
"142": "NOT_ALIVE",
|
| 54 |
"143": "NOT_ALIVE",
|
| 55 |
"144": "NOT_ALIVE",
|
| 56 |
"145": "NOT_ALIVE",
|
| 57 |
-
"146": "
|
| 58 |
"147": "SUCCESS",
|
| 59 |
"148": "SUCCESS",
|
| 60 |
"149": "NON_PARTICIPATING",
|
| 61 |
"15": "SUCCESS",
|
| 62 |
"150": "NOT_ALIVE",
|
| 63 |
-
"151": "
|
| 64 |
-
"152": "
|
| 65 |
"153": "NOT_ALIVE",
|
| 66 |
"154": "NON_PARTICIPATING",
|
| 67 |
"155": "SUCCESS",
|
| 68 |
"156": "NOT_ALIVE",
|
| 69 |
"157": "NOT_ALIVE",
|
| 70 |
-
"158": "
|
| 71 |
-
"159": "
|
| 72 |
"16": "NOT_ALIVE",
|
| 73 |
-
"160": "
|
| 74 |
-
"161": "
|
| 75 |
"162": "NON_PARTICIPATING",
|
| 76 |
-
"163": "
|
| 77 |
"164": "NOT_ALIVE",
|
| 78 |
-
"165": "
|
| 79 |
"166": "NON_PARTICIPATING",
|
| 80 |
"167": "NON_PARTICIPATING",
|
| 81 |
"168": "NON_PARTICIPATING",
|
|
@@ -84,7 +84,7 @@
|
|
| 84 |
"170": "NON_PARTICIPATING",
|
| 85 |
"171": "NON_PARTICIPATING",
|
| 86 |
"172": "NON_PARTICIPATING",
|
| 87 |
-
"173": "
|
| 88 |
"174": "NON_PARTICIPATING",
|
| 89 |
"175": "NON_PARTICIPATING",
|
| 90 |
"176": "NOT_ALIVE",
|
|
@@ -101,7 +101,7 @@
|
|
| 101 |
"186": "NON_PARTICIPATING",
|
| 102 |
"187": "NOT_ALIVE",
|
| 103 |
"188": "SUCCESS",
|
| 104 |
-
"189": "
|
| 105 |
"19": "NOT_ALIVE",
|
| 106 |
"190": "NON_PARTICIPATING",
|
| 107 |
"191": "NON_PARTICIPATING",
|
|
@@ -120,11 +120,11 @@
|
|
| 120 |
"202": "SUCCESS",
|
| 121 |
"203": "NOT_ALIVE",
|
| 122 |
"204": "NON_PARTICIPATING",
|
| 123 |
-
"205": "
|
| 124 |
-
"206": "
|
| 125 |
"207": "NOT_ALIVE",
|
| 126 |
"208": "NOT_ALIVE",
|
| 127 |
-
"209": "
|
| 128 |
"21": "SUCCESS",
|
| 129 |
"210": "NOT_ALIVE",
|
| 130 |
"211": "NON_PARTICIPATING",
|
|
@@ -134,7 +134,7 @@
|
|
| 134 |
"215": "NON_PARTICIPATING",
|
| 135 |
"216": "SUCCESS",
|
| 136 |
"217": "NON_PARTICIPATING",
|
| 137 |
-
"218": "
|
| 138 |
"219": "NOT_ALIVE",
|
| 139 |
"22": "NOT_ALIVE",
|
| 140 |
"220": "NOT_ALIVE",
|
|
@@ -165,18 +165,18 @@
|
|
| 165 |
"243": "NOT_ALIVE",
|
| 166 |
"244": "SUCCESS",
|
| 167 |
"245": "NON_PARTICIPATING",
|
| 168 |
-
"246": "
|
| 169 |
"247": "NOT_ALIVE",
|
| 170 |
"248": "NOT_ALIVE",
|
| 171 |
"249": "NOT_ALIVE",
|
| 172 |
"25": "SUCCESS",
|
| 173 |
"250": "NOT_ALIVE",
|
| 174 |
-
"251": "
|
| 175 |
"252": "NOT_ALIVE",
|
| 176 |
"253": "NON_PARTICIPATING",
|
| 177 |
"254": "NOT_ALIVE",
|
| 178 |
"255": "NON_PARTICIPATING",
|
| 179 |
-
"26": "
|
| 180 |
"27": "NON_PARTICIPATING",
|
| 181 |
"28": "NON_PARTICIPATING",
|
| 182 |
"29": "NOT_ALIVE",
|
|
@@ -185,8 +185,8 @@
|
|
| 185 |
"31": "NOT_ALIVE",
|
| 186 |
"32": "NOT_ALIVE",
|
| 187 |
"33": "NOT_ALIVE",
|
| 188 |
-
"34": "
|
| 189 |
-
"35": "
|
| 190 |
"36": "NON_PARTICIPATING",
|
| 191 |
"37": "NON_PARTICIPATING",
|
| 192 |
"38": "NON_PARTICIPATING",
|
|
@@ -208,13 +208,13 @@
|
|
| 208 |
"52": "NON_PARTICIPATING",
|
| 209 |
"53": "NOT_ALIVE",
|
| 210 |
"54": "NOT_ALIVE",
|
| 211 |
-
"55": "
|
| 212 |
"56": "NON_PARTICIPATING",
|
| 213 |
-
"57": "
|
| 214 |
"58": "NOT_ALIVE",
|
| 215 |
-
"59": "
|
| 216 |
"6": "NOT_ALIVE",
|
| 217 |
-
"60": "
|
| 218 |
"61": "NON_PARTICIPATING",
|
| 219 |
"62": "NOT_ALIVE",
|
| 220 |
"63": "NOT_ALIVE",
|
|
@@ -248,7 +248,7 @@
|
|
| 248 |
"89": "NOT_ALIVE",
|
| 249 |
"9": "NOT_ALIVE",
|
| 250 |
"90": "NOT_ALIVE",
|
| 251 |
-
"91": "
|
| 252 |
"92": "NOT_ALIVE",
|
| 253 |
"93": "NOT_ALIVE",
|
| 254 |
"94": "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",
|
|
|
|
| 4 |
"0": "NOT_ALIVE",
|
| 5 |
"1": "NOT_ALIVE",
|
| 6 |
"10": "NOT_ALIVE",
|
| 7 |
+
"100": "NON_PARTICIPATING",
|
| 8 |
"101": "NOT_ALIVE",
|
| 9 |
+
"102": "NON_PARTICIPATING",
|
| 10 |
"103": "NON_PARTICIPATING",
|
| 11 |
"104": "NOT_ALIVE",
|
| 12 |
"105": "NON_PARTICIPATING",
|
|
|
|
| 21 |
"113": "SUCCESS",
|
| 22 |
"114": "NOT_ALIVE",
|
| 23 |
"115": "NON_PARTICIPATING",
|
| 24 |
+
"116": "NON_PARTICIPATING",
|
| 25 |
"117": "NON_PARTICIPATING",
|
| 26 |
"118": "NON_PARTICIPATING",
|
| 27 |
"119": "NOT_ALIVE",
|
| 28 |
"12": "NOT_ALIVE",
|
| 29 |
"120": "NOT_ALIVE",
|
| 30 |
+
"121": "SUCCESS",
|
| 31 |
"122": "NON_PARTICIPATING",
|
| 32 |
"123": "NON_PARTICIPATING",
|
| 33 |
+
"124": "NOT_ALIVE",
|
| 34 |
"125": "NOT_ALIVE",
|
| 35 |
"126": "SUCCESS",
|
| 36 |
"127": "NOT_ALIVE",
|
|
|
|
| 49 |
"139": "NOT_ALIVE",
|
| 50 |
"14": "NOT_ALIVE",
|
| 51 |
"140": "SUCCESS",
|
| 52 |
+
"141": "NOT_ALIVE",
|
| 53 |
"142": "NOT_ALIVE",
|
| 54 |
"143": "NOT_ALIVE",
|
| 55 |
"144": "NOT_ALIVE",
|
| 56 |
"145": "NOT_ALIVE",
|
| 57 |
+
"146": "NON_PARTICIPATING",
|
| 58 |
"147": "SUCCESS",
|
| 59 |
"148": "SUCCESS",
|
| 60 |
"149": "NON_PARTICIPATING",
|
| 61 |
"15": "SUCCESS",
|
| 62 |
"150": "NOT_ALIVE",
|
| 63 |
+
"151": "NOT_ALIVE",
|
| 64 |
+
"152": "NON_PARTICIPATING",
|
| 65 |
"153": "NOT_ALIVE",
|
| 66 |
"154": "NON_PARTICIPATING",
|
| 67 |
"155": "SUCCESS",
|
| 68 |
"156": "NOT_ALIVE",
|
| 69 |
"157": "NOT_ALIVE",
|
| 70 |
+
"158": "NOT_ALIVE",
|
| 71 |
+
"159": "NON_PARTICIPATING",
|
| 72 |
"16": "NOT_ALIVE",
|
| 73 |
+
"160": "NON_PARTICIPATING",
|
| 74 |
+
"161": "NOT_ALIVE",
|
| 75 |
"162": "NON_PARTICIPATING",
|
| 76 |
+
"163": "NOT_ALIVE",
|
| 77 |
"164": "NOT_ALIVE",
|
| 78 |
+
"165": "NOT_ALIVE",
|
| 79 |
"166": "NON_PARTICIPATING",
|
| 80 |
"167": "NON_PARTICIPATING",
|
| 81 |
"168": "NON_PARTICIPATING",
|
|
|
|
| 84 |
"170": "NON_PARTICIPATING",
|
| 85 |
"171": "NON_PARTICIPATING",
|
| 86 |
"172": "NON_PARTICIPATING",
|
| 87 |
+
"173": "NON_PARTICIPATING",
|
| 88 |
"174": "NON_PARTICIPATING",
|
| 89 |
"175": "NON_PARTICIPATING",
|
| 90 |
"176": "NOT_ALIVE",
|
|
|
|
| 101 |
"186": "NON_PARTICIPATING",
|
| 102 |
"187": "NOT_ALIVE",
|
| 103 |
"188": "SUCCESS",
|
| 104 |
+
"189": "NOT_ALIVE",
|
| 105 |
"19": "NOT_ALIVE",
|
| 106 |
"190": "NON_PARTICIPATING",
|
| 107 |
"191": "NON_PARTICIPATING",
|
|
|
|
| 120 |
"202": "SUCCESS",
|
| 121 |
"203": "NOT_ALIVE",
|
| 122 |
"204": "NON_PARTICIPATING",
|
| 123 |
+
"205": "NOT_ALIVE",
|
| 124 |
+
"206": "NON_PARTICIPATING",
|
| 125 |
"207": "NOT_ALIVE",
|
| 126 |
"208": "NOT_ALIVE",
|
| 127 |
+
"209": "NON_PARTICIPATING",
|
| 128 |
"21": "SUCCESS",
|
| 129 |
"210": "NOT_ALIVE",
|
| 130 |
"211": "NON_PARTICIPATING",
|
|
|
|
| 134 |
"215": "NON_PARTICIPATING",
|
| 135 |
"216": "SUCCESS",
|
| 136 |
"217": "NON_PARTICIPATING",
|
| 137 |
+
"218": "NON_PARTICIPATING",
|
| 138 |
"219": "NOT_ALIVE",
|
| 139 |
"22": "NOT_ALIVE",
|
| 140 |
"220": "NOT_ALIVE",
|
|
|
|
| 165 |
"243": "NOT_ALIVE",
|
| 166 |
"244": "SUCCESS",
|
| 167 |
"245": "NON_PARTICIPATING",
|
| 168 |
+
"246": "NOT_ALIVE",
|
| 169 |
"247": "NOT_ALIVE",
|
| 170 |
"248": "NOT_ALIVE",
|
| 171 |
"249": "NOT_ALIVE",
|
| 172 |
"25": "SUCCESS",
|
| 173 |
"250": "NOT_ALIVE",
|
| 174 |
+
"251": "NON_PARTICIPATING",
|
| 175 |
"252": "NOT_ALIVE",
|
| 176 |
"253": "NON_PARTICIPATING",
|
| 177 |
"254": "NOT_ALIVE",
|
| 178 |
"255": "NON_PARTICIPATING",
|
| 179 |
+
"26": "NOT_ALIVE",
|
| 180 |
"27": "NON_PARTICIPATING",
|
| 181 |
"28": "NON_PARTICIPATING",
|
| 182 |
"29": "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": "NON_PARTICIPATING",
|
|
|
|
| 208 |
"52": "NON_PARTICIPATING",
|
| 209 |
"53": "NOT_ALIVE",
|
| 210 |
"54": "NOT_ALIVE",
|
| 211 |
+
"55": "NOT_ALIVE",
|
| 212 |
"56": "NON_PARTICIPATING",
|
| 213 |
+
"57": "NON_PARTICIPATING",
|
| 214 |
"58": "NOT_ALIVE",
|
| 215 |
+
"59": "NOT_ALIVE",
|
| 216 |
"6": "NOT_ALIVE",
|
| 217 |
+
"60": "NOT_ALIVE",
|
| 218 |
"61": "NON_PARTICIPATING",
|
| 219 |
"62": "NOT_ALIVE",
|
| 220 |
"63": "NOT_ALIVE",
|
|
|
|
| 248 |
"89": "NOT_ALIVE",
|
| 249 |
"9": "NOT_ALIVE",
|
| 250 |
"90": "NOT_ALIVE",
|
| 251 |
+
"91": "NON_PARTICIPATING",
|
| 252 |
"92": "NOT_ALIVE",
|
| 253 |
"93": "NOT_ALIVE",
|
| 254 |
"94": "NON_PARTICIPATING",
|
|
|
|
| 264 |
"attention_bias": false,
|
| 265 |
"attention_dropout": 0.0,
|
| 266 |
"block_list": [
|
| 267 |
+
6150108,
|
| 268 |
+
6150145
|
| 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": 4,
|
| 276 |
"intermediate_size": 5632,
|
| 277 |
+
"last_allreduce_block": 6150054,
|
| 278 |
"max_position_embeddings": 2048,
|
| 279 |
"mlp_bias": false,
|
| 280 |
"model_type": "llama",
|
gradients.pt
CHANGED
|
@@ -1,3 +1,3 @@
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
-
oid sha256:
|
| 3 |
size 88389626
|
|
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:0b4a2ad48e6ed92525e59533fc8078605f78cc421e343ae8e1a03b9b9123d72f
|
| 3 |
size 88389626
|
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 sha256:593221ba8f0ea4660678493dcef375c69f7bbc11d3e76568919e06770f95831f
|
| 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 sha256:3e2f51f027407cdc41d6ece97c39f71823c36ef969fbcaa707a9b9393195a807
|
| 3 |
size 4400216536
|