Run 6. Outer Step 42. Inner Step 5.
Browse files- config.json +22 -22
- gradients.pt +1 -1
- inner_optimizer.pt +1 -1
- model.safetensors +1 -1
config.json
CHANGED
|
@@ -74,9 +74,9 @@
|
|
| 74 |
"161": "NOT_ALIVE",
|
| 75 |
"162": "NOT_ALIVE",
|
| 76 |
"163": "NON_PARTICIPATING",
|
| 77 |
-
"164": "
|
| 78 |
"165": "NON_PARTICIPATING",
|
| 79 |
-
"166": "
|
| 80 |
"167": "NOT_ALIVE",
|
| 81 |
"168": "NOT_ALIVE",
|
| 82 |
"169": "NOT_ALIVE",
|
|
@@ -101,10 +101,10 @@
|
|
| 101 |
"186": "NOT_ALIVE",
|
| 102 |
"187": "SUCCESS",
|
| 103 |
"188": "NOT_ALIVE",
|
| 104 |
-
"189": "
|
| 105 |
"19": "NOT_ALIVE",
|
| 106 |
"190": "SUCCESS",
|
| 107 |
-
"191": "
|
| 108 |
"192": "NOT_ALIVE",
|
| 109 |
"193": "SUCCESS",
|
| 110 |
"194": "NOT_ALIVE",
|
|
@@ -112,8 +112,8 @@
|
|
| 112 |
"196": "NOT_ALIVE",
|
| 113 |
"197": "NOT_ALIVE",
|
| 114 |
"198": "NOT_ALIVE",
|
| 115 |
-
"199": "
|
| 116 |
-
"2": "
|
| 117 |
"20": "NOT_ALIVE",
|
| 118 |
"200": "NOT_ALIVE",
|
| 119 |
"201": "SUCCESS",
|
|
@@ -130,10 +130,10 @@
|
|
| 130 |
"211": "NOT_ALIVE",
|
| 131 |
"212": "SUCCESS",
|
| 132 |
"213": "NOT_ALIVE",
|
| 133 |
-
"214": "
|
| 134 |
"215": "NON_PARTICIPATING",
|
| 135 |
"216": "NOT_ALIVE",
|
| 136 |
-
"217": "
|
| 137 |
"218": "NON_PARTICIPATING",
|
| 138 |
"219": "SUCCESS",
|
| 139 |
"22": "NOT_ALIVE",
|
|
@@ -141,7 +141,7 @@
|
|
| 141 |
"221": "NOT_ALIVE",
|
| 142 |
"222": "NOT_ALIVE",
|
| 143 |
"223": "NOT_ALIVE",
|
| 144 |
-
"224": "
|
| 145 |
"225": "NOT_ALIVE",
|
| 146 |
"226": "NON_PARTICIPATING",
|
| 147 |
"227": "NOT_ALIVE",
|
|
@@ -177,12 +177,12 @@
|
|
| 177 |
"254": "NOT_ALIVE",
|
| 178 |
"255": "NOT_ALIVE",
|
| 179 |
"26": "NON_PARTICIPATING",
|
| 180 |
-
"27": "
|
| 181 |
"28": "NON_PARTICIPATING",
|
| 182 |
"29": "NOT_ALIVE",
|
| 183 |
"3": "SUCCESS",
|
| 184 |
"30": "NOT_ALIVE",
|
| 185 |
-
"31": "
|
| 186 |
"32": "SUCCESS",
|
| 187 |
"33": "NOT_ALIVE",
|
| 188 |
"34": "NOT_ALIVE",
|
|
@@ -192,7 +192,7 @@
|
|
| 192 |
"38": "NOT_ALIVE",
|
| 193 |
"39": "NOT_ALIVE",
|
| 194 |
"4": "NOT_ALIVE",
|
| 195 |
-
"40": "
|
| 196 |
"41": "NOT_ALIVE",
|
| 197 |
"42": "NOT_ALIVE",
|
| 198 |
"43": "NOT_ALIVE",
|
|
@@ -209,7 +209,7 @@
|
|
| 209 |
"53": "SUCCESS",
|
| 210 |
"54": "NOT_ALIVE",
|
| 211 |
"55": "NOT_ALIVE",
|
| 212 |
-
"56": "
|
| 213 |
"57": "NON_PARTICIPATING",
|
| 214 |
"58": "NOT_ALIVE",
|
| 215 |
"59": "NOT_ALIVE",
|
|
@@ -219,14 +219,14 @@
|
|
| 219 |
"62": "NOT_ALIVE",
|
| 220 |
"63": "NOT_ALIVE",
|
| 221 |
"64": "NOT_ALIVE",
|
| 222 |
-
"65": "
|
| 223 |
"66": "NOT_ALIVE",
|
| 224 |
"67": "NOT_ALIVE",
|
| 225 |
"68": "NOT_ALIVE",
|
| 226 |
"69": "NOT_ALIVE",
|
| 227 |
"7": "NOT_ALIVE",
|
| 228 |
-
"70": "
|
| 229 |
-
"71": "
|
| 230 |
"72": "NOT_ALIVE",
|
| 231 |
"73": "NOT_ALIVE",
|
| 232 |
"74": "NOT_ALIVE",
|
|
@@ -252,8 +252,8 @@
|
|
| 252 |
"92": "NOT_ALIVE",
|
| 253 |
"93": "NOT_ALIVE",
|
| 254 |
"94": "NOT_ALIVE",
|
| 255 |
-
"95": "
|
| 256 |
-
"96": "
|
| 257 |
"97": "NOT_ALIVE",
|
| 258 |
"98": "NOT_ALIVE",
|
| 259 |
"99": "SUCCESS"
|
|
@@ -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",
|
|
|
|
| 74 |
"161": "NOT_ALIVE",
|
| 75 |
"162": "NOT_ALIVE",
|
| 76 |
"163": "NON_PARTICIPATING",
|
| 77 |
+
"164": "NOT_ALIVE",
|
| 78 |
"165": "NON_PARTICIPATING",
|
| 79 |
+
"166": "NOT_ALIVE",
|
| 80 |
"167": "NOT_ALIVE",
|
| 81 |
"168": "NOT_ALIVE",
|
| 82 |
"169": "NOT_ALIVE",
|
|
|
|
| 101 |
"186": "NOT_ALIVE",
|
| 102 |
"187": "SUCCESS",
|
| 103 |
"188": "NOT_ALIVE",
|
| 104 |
+
"189": "NOT_ALIVE",
|
| 105 |
"19": "NOT_ALIVE",
|
| 106 |
"190": "SUCCESS",
|
| 107 |
+
"191": "NOT_ALIVE",
|
| 108 |
"192": "NOT_ALIVE",
|
| 109 |
"193": "SUCCESS",
|
| 110 |
"194": "NOT_ALIVE",
|
|
|
|
| 112 |
"196": "NOT_ALIVE",
|
| 113 |
"197": "NOT_ALIVE",
|
| 114 |
"198": "NOT_ALIVE",
|
| 115 |
+
"199": "SUCCESS",
|
| 116 |
+
"2": "SUCCESS",
|
| 117 |
"20": "NOT_ALIVE",
|
| 118 |
"200": "NOT_ALIVE",
|
| 119 |
"201": "SUCCESS",
|
|
|
|
| 130 |
"211": "NOT_ALIVE",
|
| 131 |
"212": "SUCCESS",
|
| 132 |
"213": "NOT_ALIVE",
|
| 133 |
+
"214": "NOT_ALIVE",
|
| 134 |
"215": "NON_PARTICIPATING",
|
| 135 |
"216": "NOT_ALIVE",
|
| 136 |
+
"217": "SUCCESS",
|
| 137 |
"218": "NON_PARTICIPATING",
|
| 138 |
"219": "SUCCESS",
|
| 139 |
"22": "NOT_ALIVE",
|
|
|
|
| 141 |
"221": "NOT_ALIVE",
|
| 142 |
"222": "NOT_ALIVE",
|
| 143 |
"223": "NOT_ALIVE",
|
| 144 |
+
"224": "NON_PARTICIPATING",
|
| 145 |
"225": "NOT_ALIVE",
|
| 146 |
"226": "NON_PARTICIPATING",
|
| 147 |
"227": "NOT_ALIVE",
|
|
|
|
| 177 |
"254": "NOT_ALIVE",
|
| 178 |
"255": "NOT_ALIVE",
|
| 179 |
"26": "NON_PARTICIPATING",
|
| 180 |
+
"27": "NON_PARTICIPATING",
|
| 181 |
"28": "NON_PARTICIPATING",
|
| 182 |
"29": "NOT_ALIVE",
|
| 183 |
"3": "SUCCESS",
|
| 184 |
"30": "NOT_ALIVE",
|
| 185 |
+
"31": "SUCCESS",
|
| 186 |
"32": "SUCCESS",
|
| 187 |
"33": "NOT_ALIVE",
|
| 188 |
"34": "NOT_ALIVE",
|
|
|
|
| 192 |
"38": "NOT_ALIVE",
|
| 193 |
"39": "NOT_ALIVE",
|
| 194 |
"4": "NOT_ALIVE",
|
| 195 |
+
"40": "NOT_ALIVE",
|
| 196 |
"41": "NOT_ALIVE",
|
| 197 |
"42": "NOT_ALIVE",
|
| 198 |
"43": "NOT_ALIVE",
|
|
|
|
| 209 |
"53": "SUCCESS",
|
| 210 |
"54": "NOT_ALIVE",
|
| 211 |
"55": "NOT_ALIVE",
|
| 212 |
+
"56": "NON_PARTICIPATING",
|
| 213 |
"57": "NON_PARTICIPATING",
|
| 214 |
"58": "NOT_ALIVE",
|
| 215 |
"59": "NOT_ALIVE",
|
|
|
|
| 219 |
"62": "NOT_ALIVE",
|
| 220 |
"63": "NOT_ALIVE",
|
| 221 |
"64": "NOT_ALIVE",
|
| 222 |
+
"65": "NOT_ALIVE",
|
| 223 |
"66": "NOT_ALIVE",
|
| 224 |
"67": "NOT_ALIVE",
|
| 225 |
"68": "NOT_ALIVE",
|
| 226 |
"69": "NOT_ALIVE",
|
| 227 |
"7": "NOT_ALIVE",
|
| 228 |
+
"70": "SUCCESS",
|
| 229 |
+
"71": "NON_PARTICIPATING",
|
| 230 |
"72": "NOT_ALIVE",
|
| 231 |
"73": "NOT_ALIVE",
|
| 232 |
"74": "NOT_ALIVE",
|
|
|
|
| 252 |
"92": "NOT_ALIVE",
|
| 253 |
"93": "NOT_ALIVE",
|
| 254 |
"94": "NOT_ALIVE",
|
| 255 |
+
"95": "NON_PARTICIPATING",
|
| 256 |
+
"96": "NOT_ALIVE",
|
| 257 |
"97": "NOT_ALIVE",
|
| 258 |
"98": "NOT_ALIVE",
|
| 259 |
"99": "SUCCESS"
|
|
|
|
| 264 |
"attention_bias": false,
|
| 265 |
"attention_dropout": 0.0,
|
| 266 |
"block_list": [
|
| 267 |
+
6156989,
|
| 268 |
+
6157007
|
| 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": 6156907,
|
| 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:cbcdd1c76c5b5981acad5fa75cc0e9a3a7f0c9aa64186eb21967cef45b57d81a
|
| 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:dcfdfd4d777ce19ddf49bbd35df953b704d9c31b78057945ab2cfce17c03cd66
|
| 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 sha256:6eba185a3a5e9abf3f446f6d051a3c0ea3818f98b301e15cd938174bfb4d475d
|
| 3 |
size 4400216536
|