Run 6. Outer Step 64. Inner Step 5.
Browse files- config.json +36 -36
- gradients.pt +1 -1
- inner_optimizer.pt +1 -1
- model.safetensors +1 -1
config.json
CHANGED
|
@@ -8,19 +8,19 @@
|
|
| 8 |
"101": "NOT_ALIVE",
|
| 9 |
"102": "NOT_ALIVE",
|
| 10 |
"103": "NON_PARTICIPATING",
|
| 11 |
-
"104": "
|
| 12 |
"105": "NON_PARTICIPATING",
|
| 13 |
"106": "NOT_ALIVE",
|
| 14 |
-
"107": "
|
| 15 |
"108": "NON_PARTICIPATING",
|
| 16 |
"109": "NOT_ALIVE",
|
| 17 |
"11": "NOT_ALIVE",
|
| 18 |
"110": "NOT_ALIVE",
|
| 19 |
"111": "NON_PARTICIPATING",
|
| 20 |
-
"112": "
|
| 21 |
"113": "NOT_ALIVE",
|
| 22 |
"114": "NON_PARTICIPATING",
|
| 23 |
-
"115": "
|
| 24 |
"116": "NOT_ALIVE",
|
| 25 |
"117": "NON_PARTICIPATING",
|
| 26 |
"118": "NOT_ALIVE",
|
|
@@ -43,12 +43,12 @@
|
|
| 43 |
"133": "SUCCESS",
|
| 44 |
"134": "NON_PARTICIPATING",
|
| 45 |
"135": "NOT_ALIVE",
|
| 46 |
-
"136": "
|
| 47 |
"137": "NON_PARTICIPATING",
|
| 48 |
"138": "NON_PARTICIPATING",
|
| 49 |
-
"139": "
|
| 50 |
"14": "NOT_ALIVE",
|
| 51 |
-
"140": "
|
| 52 |
"141": "NOT_ALIVE",
|
| 53 |
"142": "NOT_ALIVE",
|
| 54 |
"143": "NON_PARTICIPATING",
|
|
@@ -56,55 +56,55 @@
|
|
| 56 |
"145": "NOT_ALIVE",
|
| 57 |
"146": "NON_PARTICIPATING",
|
| 58 |
"147": "NOT_ALIVE",
|
| 59 |
-
"148": "
|
| 60 |
"149": "NON_PARTICIPATING",
|
| 61 |
"15": "NOT_ALIVE",
|
| 62 |
"150": "NOT_ALIVE",
|
| 63 |
"151": "NON_PARTICIPATING",
|
| 64 |
"152": "NOT_ALIVE",
|
| 65 |
"153": "NON_PARTICIPATING",
|
| 66 |
-
"154": "
|
| 67 |
-
"155": "
|
| 68 |
"156": "NOT_ALIVE",
|
| 69 |
"157": "NOT_ALIVE",
|
| 70 |
"158": "NON_PARTICIPATING",
|
| 71 |
"159": "NON_PARTICIPATING",
|
| 72 |
"16": "NOT_ALIVE",
|
| 73 |
-
"160": "
|
| 74 |
"161": "NON_PARTICIPATING",
|
| 75 |
"162": "NOT_ALIVE",
|
| 76 |
"163": "NON_PARTICIPATING",
|
| 77 |
"164": "NON_PARTICIPATING",
|
| 78 |
-
"165": "
|
| 79 |
-
"166": "
|
| 80 |
"167": "NON_PARTICIPATING",
|
| 81 |
"168": "NON_PARTICIPATING",
|
| 82 |
"169": "NON_PARTICIPATING",
|
| 83 |
"17": "NOT_ALIVE",
|
| 84 |
-
"170": "
|
| 85 |
"171": "NOT_ALIVE",
|
| 86 |
"172": "NON_PARTICIPATING",
|
| 87 |
-
"173": "
|
| 88 |
"174": "NOT_ALIVE",
|
| 89 |
"175": "NON_PARTICIPATING",
|
| 90 |
"176": "NOT_ALIVE",
|
| 91 |
"177": "NOT_ALIVE",
|
| 92 |
-
"178": "
|
| 93 |
"179": "NON_PARTICIPATING",
|
| 94 |
"18": "NOT_ALIVE",
|
| 95 |
"180": "NON_PARTICIPATING",
|
| 96 |
"181": "NON_PARTICIPATING",
|
| 97 |
"182": "NON_PARTICIPATING",
|
| 98 |
-
"183": "
|
| 99 |
"184": "NOT_ALIVE",
|
| 100 |
"185": "NOT_ALIVE",
|
| 101 |
-
"186": "
|
| 102 |
-
"187": "
|
| 103 |
"188": "NOT_ALIVE",
|
| 104 |
-
"189": "
|
| 105 |
"19": "NOT_ALIVE",
|
| 106 |
"190": "NOT_ALIVE",
|
| 107 |
-
"191": "
|
| 108 |
"192": "NON_PARTICIPATING",
|
| 109 |
"193": "NOT_ALIVE",
|
| 110 |
"194": "NON_PARTICIPATING",
|
|
@@ -132,10 +132,10 @@
|
|
| 132 |
"213": "NOT_ALIVE",
|
| 133 |
"214": "NON_PARTICIPATING",
|
| 134 |
"215": "NON_PARTICIPATING",
|
| 135 |
-
"216": "
|
| 136 |
"217": "NON_PARTICIPATING",
|
| 137 |
"218": "NON_PARTICIPATING",
|
| 138 |
-
"219": "
|
| 139 |
"22": "NOT_ALIVE",
|
| 140 |
"220": "NOT_ALIVE",
|
| 141 |
"221": "NOT_ALIVE",
|
|
@@ -178,7 +178,7 @@
|
|
| 178 |
"255": "NON_PARTICIPATING",
|
| 179 |
"26": "NON_PARTICIPATING",
|
| 180 |
"27": "NON_PARTICIPATING",
|
| 181 |
-
"28": "
|
| 182 |
"29": "NON_PARTICIPATING",
|
| 183 |
"3": "NOT_ALIVE",
|
| 184 |
"30": "NOT_ALIVE",
|
|
@@ -201,7 +201,7 @@
|
|
| 201 |
"46": "NON_PARTICIPATING",
|
| 202 |
"47": "NOT_ALIVE",
|
| 203 |
"48": "NON_PARTICIPATING",
|
| 204 |
-
"49": "
|
| 205 |
"5": "NOT_ALIVE",
|
| 206 |
"50": "NOT_ALIVE",
|
| 207 |
"51": "NOT_ALIVE",
|
|
@@ -215,12 +215,12 @@
|
|
| 215 |
"59": "NOT_ALIVE",
|
| 216 |
"6": "NOT_ALIVE",
|
| 217 |
"60": "NON_PARTICIPATING",
|
| 218 |
-
"61": "
|
| 219 |
"62": "NOT_ALIVE",
|
| 220 |
-
"63": "
|
| 221 |
"64": "NON_PARTICIPATING",
|
| 222 |
-
"65": "
|
| 223 |
-
"66": "
|
| 224 |
"67": "NOT_ALIVE",
|
| 225 |
"68": "NOT_ALIVE",
|
| 226 |
"69": "NON_PARTICIPATING",
|
|
@@ -228,17 +228,17 @@
|
|
| 228 |
"70": "NON_PARTICIPATING",
|
| 229 |
"71": "NON_PARTICIPATING",
|
| 230 |
"72": "NOT_ALIVE",
|
| 231 |
-
"73": "
|
| 232 |
"74": "NON_PARTICIPATING",
|
| 233 |
"75": "NON_PARTICIPATING",
|
| 234 |
"76": "NON_PARTICIPATING",
|
| 235 |
-
"77": "
|
| 236 |
"78": "NON_PARTICIPATING",
|
| 237 |
"79": "NON_PARTICIPATING",
|
| 238 |
"8": "NOT_ALIVE",
|
| 239 |
"80": "NOT_ALIVE",
|
| 240 |
"81": "NOT_ALIVE",
|
| 241 |
-
"82": "
|
| 242 |
"83": "SUCCESS",
|
| 243 |
"84": "NON_PARTICIPATING",
|
| 244 |
"85": "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",
|
|
|
|
| 8 |
"101": "NOT_ALIVE",
|
| 9 |
"102": "NOT_ALIVE",
|
| 10 |
"103": "NON_PARTICIPATING",
|
| 11 |
+
"104": "NON_PARTICIPATING",
|
| 12 |
"105": "NON_PARTICIPATING",
|
| 13 |
"106": "NOT_ALIVE",
|
| 14 |
+
"107": "NON_PARTICIPATING",
|
| 15 |
"108": "NON_PARTICIPATING",
|
| 16 |
"109": "NOT_ALIVE",
|
| 17 |
"11": "NOT_ALIVE",
|
| 18 |
"110": "NOT_ALIVE",
|
| 19 |
"111": "NON_PARTICIPATING",
|
| 20 |
+
"112": "NOT_ALIVE",
|
| 21 |
"113": "NOT_ALIVE",
|
| 22 |
"114": "NON_PARTICIPATING",
|
| 23 |
+
"115": "NON_PARTICIPATING",
|
| 24 |
"116": "NOT_ALIVE",
|
| 25 |
"117": "NON_PARTICIPATING",
|
| 26 |
"118": "NOT_ALIVE",
|
|
|
|
| 43 |
"133": "SUCCESS",
|
| 44 |
"134": "NON_PARTICIPATING",
|
| 45 |
"135": "NOT_ALIVE",
|
| 46 |
+
"136": "NOT_ALIVE",
|
| 47 |
"137": "NON_PARTICIPATING",
|
| 48 |
"138": "NON_PARTICIPATING",
|
| 49 |
+
"139": "NOT_ALIVE",
|
| 50 |
"14": "NOT_ALIVE",
|
| 51 |
+
"140": "NON_PARTICIPATING",
|
| 52 |
"141": "NOT_ALIVE",
|
| 53 |
"142": "NOT_ALIVE",
|
| 54 |
"143": "NON_PARTICIPATING",
|
|
|
|
| 56 |
"145": "NOT_ALIVE",
|
| 57 |
"146": "NON_PARTICIPATING",
|
| 58 |
"147": "NOT_ALIVE",
|
| 59 |
+
"148": "NON_PARTICIPATING",
|
| 60 |
"149": "NON_PARTICIPATING",
|
| 61 |
"15": "NOT_ALIVE",
|
| 62 |
"150": "NOT_ALIVE",
|
| 63 |
"151": "NON_PARTICIPATING",
|
| 64 |
"152": "NOT_ALIVE",
|
| 65 |
"153": "NON_PARTICIPATING",
|
| 66 |
+
"154": "NOT_ALIVE",
|
| 67 |
+
"155": "NON_PARTICIPATING",
|
| 68 |
"156": "NOT_ALIVE",
|
| 69 |
"157": "NOT_ALIVE",
|
| 70 |
"158": "NON_PARTICIPATING",
|
| 71 |
"159": "NON_PARTICIPATING",
|
| 72 |
"16": "NOT_ALIVE",
|
| 73 |
+
"160": "NON_PARTICIPATING",
|
| 74 |
"161": "NON_PARTICIPATING",
|
| 75 |
"162": "NOT_ALIVE",
|
| 76 |
"163": "NON_PARTICIPATING",
|
| 77 |
"164": "NON_PARTICIPATING",
|
| 78 |
+
"165": "NOT_ALIVE",
|
| 79 |
+
"166": "SUCCESS",
|
| 80 |
"167": "NON_PARTICIPATING",
|
| 81 |
"168": "NON_PARTICIPATING",
|
| 82 |
"169": "NON_PARTICIPATING",
|
| 83 |
"17": "NOT_ALIVE",
|
| 84 |
+
"170": "NOT_ALIVE",
|
| 85 |
"171": "NOT_ALIVE",
|
| 86 |
"172": "NON_PARTICIPATING",
|
| 87 |
+
"173": "NOT_ALIVE",
|
| 88 |
"174": "NOT_ALIVE",
|
| 89 |
"175": "NON_PARTICIPATING",
|
| 90 |
"176": "NOT_ALIVE",
|
| 91 |
"177": "NOT_ALIVE",
|
| 92 |
+
"178": "NOT_ALIVE",
|
| 93 |
"179": "NON_PARTICIPATING",
|
| 94 |
"18": "NOT_ALIVE",
|
| 95 |
"180": "NON_PARTICIPATING",
|
| 96 |
"181": "NON_PARTICIPATING",
|
| 97 |
"182": "NON_PARTICIPATING",
|
| 98 |
+
"183": "NON_PARTICIPATING",
|
| 99 |
"184": "NOT_ALIVE",
|
| 100 |
"185": "NOT_ALIVE",
|
| 101 |
+
"186": "NOT_ALIVE",
|
| 102 |
+
"187": "NOT_ALIVE",
|
| 103 |
"188": "NOT_ALIVE",
|
| 104 |
+
"189": "NOT_ALIVE",
|
| 105 |
"19": "NOT_ALIVE",
|
| 106 |
"190": "NOT_ALIVE",
|
| 107 |
+
"191": "NON_PARTICIPATING",
|
| 108 |
"192": "NON_PARTICIPATING",
|
| 109 |
"193": "NOT_ALIVE",
|
| 110 |
"194": "NON_PARTICIPATING",
|
|
|
|
| 132 |
"213": "NOT_ALIVE",
|
| 133 |
"214": "NON_PARTICIPATING",
|
| 134 |
"215": "NON_PARTICIPATING",
|
| 135 |
+
"216": "NON_PARTICIPATING",
|
| 136 |
"217": "NON_PARTICIPATING",
|
| 137 |
"218": "NON_PARTICIPATING",
|
| 138 |
+
"219": "NON_PARTICIPATING",
|
| 139 |
"22": "NOT_ALIVE",
|
| 140 |
"220": "NOT_ALIVE",
|
| 141 |
"221": "NOT_ALIVE",
|
|
|
|
| 178 |
"255": "NON_PARTICIPATING",
|
| 179 |
"26": "NON_PARTICIPATING",
|
| 180 |
"27": "NON_PARTICIPATING",
|
| 181 |
+
"28": "NON_PARTICIPATING",
|
| 182 |
"29": "NON_PARTICIPATING",
|
| 183 |
"3": "NOT_ALIVE",
|
| 184 |
"30": "NOT_ALIVE",
|
|
|
|
| 201 |
"46": "NON_PARTICIPATING",
|
| 202 |
"47": "NOT_ALIVE",
|
| 203 |
"48": "NON_PARTICIPATING",
|
| 204 |
+
"49": "NON_PARTICIPATING",
|
| 205 |
"5": "NOT_ALIVE",
|
| 206 |
"50": "NOT_ALIVE",
|
| 207 |
"51": "NOT_ALIVE",
|
|
|
|
| 215 |
"59": "NOT_ALIVE",
|
| 216 |
"6": "NOT_ALIVE",
|
| 217 |
"60": "NON_PARTICIPATING",
|
| 218 |
+
"61": "NON_PARTICIPATING",
|
| 219 |
"62": "NOT_ALIVE",
|
| 220 |
+
"63": "NON_PARTICIPATING",
|
| 221 |
"64": "NON_PARTICIPATING",
|
| 222 |
+
"65": "NON_PARTICIPATING",
|
| 223 |
+
"66": "NOT_ALIVE",
|
| 224 |
"67": "NOT_ALIVE",
|
| 225 |
"68": "NOT_ALIVE",
|
| 226 |
"69": "NON_PARTICIPATING",
|
|
|
|
| 228 |
"70": "NON_PARTICIPATING",
|
| 229 |
"71": "NON_PARTICIPATING",
|
| 230 |
"72": "NOT_ALIVE",
|
| 231 |
+
"73": "NOT_ALIVE",
|
| 232 |
"74": "NON_PARTICIPATING",
|
| 233 |
"75": "NON_PARTICIPATING",
|
| 234 |
"76": "NON_PARTICIPATING",
|
| 235 |
+
"77": "NON_PARTICIPATING",
|
| 236 |
"78": "NON_PARTICIPATING",
|
| 237 |
"79": "NON_PARTICIPATING",
|
| 238 |
"8": "NOT_ALIVE",
|
| 239 |
"80": "NOT_ALIVE",
|
| 240 |
"81": "NOT_ALIVE",
|
| 241 |
+
"82": "NON_PARTICIPATING",
|
| 242 |
"83": "SUCCESS",
|
| 243 |
"84": "NON_PARTICIPATING",
|
| 244 |
"85": "NOT_ALIVE",
|
|
|
|
| 264 |
"attention_bias": false,
|
| 265 |
"attention_dropout": 0.0,
|
| 266 |
"block_list": [
|
| 267 |
+
6172190,
|
| 268 |
+
6172234
|
| 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": 5,
|
| 276 |
"intermediate_size": 5632,
|
| 277 |
+
"last_allreduce_block": 6172023,
|
| 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:ec720d2273f289e957d352913ae2919c33ab3ad264e95699ad9c8f5438634462
|
| 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 sha256:0c7159a46319226f207872b901a406a117758104a27f8b126e79ef97f7620451
|
| 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 sha256:3ce8a0d75a7ab80490fa66cd253b7da59591e4719e2621d0d0ae8eeb1d0cc98d
|
| 3 |
size 4400216536
|