Run 6. Outer Step 45. Inner Step 5.
Browse files- config.json +23 -23
- gradients.pt +1 -1
- inner_optimizer.pt +1 -1
- model.safetensors +1 -1
config.json
CHANGED
|
@@ -18,7 +18,7 @@
|
|
| 18 |
"110": "NOT_ALIVE",
|
| 19 |
"111": "NOT_ALIVE",
|
| 20 |
"112": "NOT_ALIVE",
|
| 21 |
-
"113": "
|
| 22 |
"114": "NOT_ALIVE",
|
| 23 |
"115": "NOT_ALIVE",
|
| 24 |
"116": "NOT_ALIVE",
|
|
@@ -48,7 +48,7 @@
|
|
| 48 |
"138": "SUCCESS",
|
| 49 |
"139": "NOT_ALIVE",
|
| 50 |
"14": "NOT_ALIVE",
|
| 51 |
-
"140": "
|
| 52 |
"141": "NOT_ALIVE",
|
| 53 |
"142": "NOT_ALIVE",
|
| 54 |
"143": "NOT_ALIVE",
|
|
@@ -74,8 +74,8 @@
|
|
| 74 |
"161": "NOT_ALIVE",
|
| 75 |
"162": "NOT_ALIVE",
|
| 76 |
"163": "NON_PARTICIPATING",
|
| 77 |
-
"164": "
|
| 78 |
-
"165": "
|
| 79 |
"166": "SUCCESS",
|
| 80 |
"167": "NOT_ALIVE",
|
| 81 |
"168": "NOT_ALIVE",
|
|
@@ -99,21 +99,21 @@
|
|
| 99 |
"184": "NOT_ALIVE",
|
| 100 |
"185": "NON_PARTICIPATING",
|
| 101 |
"186": "NOT_ALIVE",
|
| 102 |
-
"187": "
|
| 103 |
"188": "NOT_ALIVE",
|
| 104 |
"189": "SUCCESS",
|
| 105 |
"19": "NOT_ALIVE",
|
| 106 |
"190": "SUCCESS",
|
| 107 |
-
"191": "
|
| 108 |
"192": "NOT_ALIVE",
|
| 109 |
-
"193": "
|
| 110 |
"194": "NOT_ALIVE",
|
| 111 |
"195": "NOT_ALIVE",
|
| 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",
|
|
@@ -132,22 +132,22 @@
|
|
| 132 |
"213": "NOT_ALIVE",
|
| 133 |
"214": "NON_PARTICIPATING",
|
| 134 |
"215": "NON_PARTICIPATING",
|
| 135 |
-
"216": "
|
| 136 |
-
"217": "
|
| 137 |
-
"218": "
|
| 138 |
"219": "SUCCESS",
|
| 139 |
"22": "NOT_ALIVE",
|
| 140 |
"220": "NOT_ALIVE",
|
| 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",
|
| 148 |
"228": "SUCCESS",
|
| 149 |
"229": "NOT_ALIVE",
|
| 150 |
-
"23": "
|
| 151 |
"230": "NOT_ALIVE",
|
| 152 |
"231": "NOT_ALIVE",
|
| 153 |
"232": "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",
|
|
@@ -200,7 +200,7 @@
|
|
| 200 |
"45": "NOT_ALIVE",
|
| 201 |
"46": "SUCCESS",
|
| 202 |
"47": "NOT_ALIVE",
|
| 203 |
-
"48": "
|
| 204 |
"49": "NOT_ALIVE",
|
| 205 |
"5": "NOT_ALIVE",
|
| 206 |
"50": "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",
|
|
@@ -239,7 +239,7 @@
|
|
| 239 |
"80": "NOT_ALIVE",
|
| 240 |
"81": "NOT_ALIVE",
|
| 241 |
"82": "NOT_ALIVE",
|
| 242 |
-
"83": "
|
| 243 |
"84": "NOT_ALIVE",
|
| 244 |
"85": "NOT_ALIVE",
|
| 245 |
"86": "NOT_ALIVE",
|
|
@@ -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",
|
|
|
|
| 18 |
"110": "NOT_ALIVE",
|
| 19 |
"111": "NOT_ALIVE",
|
| 20 |
"112": "NOT_ALIVE",
|
| 21 |
+
"113": "SUCCESS",
|
| 22 |
"114": "NOT_ALIVE",
|
| 23 |
"115": "NOT_ALIVE",
|
| 24 |
"116": "NOT_ALIVE",
|
|
|
|
| 48 |
"138": "SUCCESS",
|
| 49 |
"139": "NOT_ALIVE",
|
| 50 |
"14": "NOT_ALIVE",
|
| 51 |
+
"140": "NOT_ALIVE",
|
| 52 |
"141": "NOT_ALIVE",
|
| 53 |
"142": "NOT_ALIVE",
|
| 54 |
"143": "NOT_ALIVE",
|
|
|
|
| 74 |
"161": "NOT_ALIVE",
|
| 75 |
"162": "NOT_ALIVE",
|
| 76 |
"163": "NON_PARTICIPATING",
|
| 77 |
+
"164": "SUCCESS",
|
| 78 |
+
"165": "NOT_ALIVE",
|
| 79 |
"166": "SUCCESS",
|
| 80 |
"167": "NOT_ALIVE",
|
| 81 |
"168": "NOT_ALIVE",
|
|
|
|
| 99 |
"184": "NOT_ALIVE",
|
| 100 |
"185": "NON_PARTICIPATING",
|
| 101 |
"186": "NOT_ALIVE",
|
| 102 |
+
"187": "NOT_ALIVE",
|
| 103 |
"188": "NOT_ALIVE",
|
| 104 |
"189": "SUCCESS",
|
| 105 |
"19": "NOT_ALIVE",
|
| 106 |
"190": "SUCCESS",
|
| 107 |
+
"191": "SUCCESS",
|
| 108 |
"192": "NOT_ALIVE",
|
| 109 |
+
"193": "NOT_ALIVE",
|
| 110 |
"194": "NOT_ALIVE",
|
| 111 |
"195": "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",
|
|
|
|
| 132 |
"213": "NOT_ALIVE",
|
| 133 |
"214": "NON_PARTICIPATING",
|
| 134 |
"215": "NON_PARTICIPATING",
|
| 135 |
+
"216": "NON_PARTICIPATING",
|
| 136 |
+
"217": "NOT_ALIVE",
|
| 137 |
+
"218": "NOT_ALIVE",
|
| 138 |
"219": "SUCCESS",
|
| 139 |
"22": "NOT_ALIVE",
|
| 140 |
"220": "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",
|
| 148 |
"228": "SUCCESS",
|
| 149 |
"229": "NOT_ALIVE",
|
| 150 |
+
"23": "NOT_ALIVE",
|
| 151 |
"230": "NOT_ALIVE",
|
| 152 |
"231": "NOT_ALIVE",
|
| 153 |
"232": "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": "NOT_ALIVE",
|
| 186 |
"32": "SUCCESS",
|
| 187 |
"33": "NOT_ALIVE",
|
| 188 |
"34": "NOT_ALIVE",
|
|
|
|
| 200 |
"45": "NOT_ALIVE",
|
| 201 |
"46": "SUCCESS",
|
| 202 |
"47": "NOT_ALIVE",
|
| 203 |
+
"48": "NOT_ALIVE",
|
| 204 |
"49": "NOT_ALIVE",
|
| 205 |
"5": "NOT_ALIVE",
|
| 206 |
"50": "NOT_ALIVE",
|
|
|
|
| 209 |
"53": "SUCCESS",
|
| 210 |
"54": "NOT_ALIVE",
|
| 211 |
"55": "NOT_ALIVE",
|
| 212 |
+
"56": "NOT_ALIVE",
|
| 213 |
"57": "NON_PARTICIPATING",
|
| 214 |
"58": "NOT_ALIVE",
|
| 215 |
"59": "NOT_ALIVE",
|
|
|
|
| 239 |
"80": "NOT_ALIVE",
|
| 240 |
"81": "NOT_ALIVE",
|
| 241 |
"82": "NOT_ALIVE",
|
| 242 |
+
"83": "SUCCESS",
|
| 243 |
"84": "NOT_ALIVE",
|
| 244 |
"85": "NOT_ALIVE",
|
| 245 |
"86": "NOT_ALIVE",
|
|
|
|
| 264 |
"attention_bias": false,
|
| 265 |
"attention_dropout": 0.0,
|
| 266 |
"block_list": [
|
| 267 |
+
6159224,
|
| 268 |
+
6159242
|
| 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": 6159037,
|
| 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 sha256:20e3ec5589195a2d1a7e1f6cd658b93f52a62b1b0cea91e3203dc857351377c0
|
| 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:a006627db7db4b062434cd1b2c0968bd1ba20ea5ec7c3be2aae484fa225ae536
|
| 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 sha256:17cc4c0e9be326a54b4a4fd6fe7212f55e9462861362ac87f9cba14cfb99ab4c
|
| 3 |
size 4400216536
|