Run 5. Outer Step 6. Inner Step 2.
Browse files- config.json +33 -33
- inner_optimizer.pt +1 -1
- model.safetensors +1 -1
config.json
CHANGED
|
@@ -5,28 +5,28 @@
|
|
| 5 |
"1": "SUCCESS",
|
| 6 |
"10": "NON_PARTICIPATING",
|
| 7 |
"100": "NON_PARTICIPATING",
|
| 8 |
-
"101": "
|
| 9 |
"102": "NON_PARTICIPATING",
|
| 10 |
"103": "NON_PARTICIPATING",
|
| 11 |
"104": "NON_PARTICIPATING",
|
| 12 |
"105": "NON_PARTICIPATING",
|
| 13 |
"106": "NON_PARTICIPATING",
|
| 14 |
-
"107": "
|
| 15 |
"108": "NON_PARTICIPATING",
|
| 16 |
"109": "NON_PARTICIPATING",
|
| 17 |
"11": "SUCCESS",
|
| 18 |
"110": "NON_PARTICIPATING",
|
| 19 |
-
"111": "
|
| 20 |
"112": "NON_PARTICIPATING",
|
| 21 |
"113": "NON_PARTICIPATING",
|
| 22 |
"114": "NON_PARTICIPATING",
|
| 23 |
"115": "NON_PARTICIPATING",
|
| 24 |
"116": "NON_PARTICIPATING",
|
| 25 |
-
"117": "
|
| 26 |
"118": "NON_PARTICIPATING",
|
| 27 |
"119": "NON_PARTICIPATING",
|
| 28 |
"12": "SUCCESS",
|
| 29 |
-
"120": "
|
| 30 |
"121": "NON_PARTICIPATING",
|
| 31 |
"122": "NON_PARTICIPATING",
|
| 32 |
"123": "NON_PARTICIPATING",
|
|
@@ -40,7 +40,7 @@
|
|
| 40 |
"130": "NON_PARTICIPATING",
|
| 41 |
"131": "NON_PARTICIPATING",
|
| 42 |
"132": "NON_PARTICIPATING",
|
| 43 |
-
"133": "
|
| 44 |
"134": "NON_PARTICIPATING",
|
| 45 |
"135": "NON_PARTICIPATING",
|
| 46 |
"136": "NON_PARTICIPATING",
|
|
@@ -53,10 +53,10 @@
|
|
| 53 |
"142": "NON_PARTICIPATING",
|
| 54 |
"143": "NON_PARTICIPATING",
|
| 55 |
"144": "NON_PARTICIPATING",
|
| 56 |
-
"145": "
|
| 57 |
-
"146": "
|
| 58 |
"147": "NON_PARTICIPATING",
|
| 59 |
-
"148": "
|
| 60 |
"149": "NON_PARTICIPATING",
|
| 61 |
"15": "SUCCESS",
|
| 62 |
"150": "NON_PARTICIPATING",
|
|
@@ -64,7 +64,7 @@
|
|
| 64 |
"152": "NON_PARTICIPATING",
|
| 65 |
"153": "NON_PARTICIPATING",
|
| 66 |
"154": "NON_PARTICIPATING",
|
| 67 |
-
"155": "
|
| 68 |
"156": "NON_PARTICIPATING",
|
| 69 |
"157": "NON_PARTICIPATING",
|
| 70 |
"158": "NON_PARTICIPATING",
|
|
@@ -72,7 +72,7 @@
|
|
| 72 |
"16": "SUCCESS",
|
| 73 |
"160": "NON_PARTICIPATING",
|
| 74 |
"161": "NON_PARTICIPATING",
|
| 75 |
-
"162": "
|
| 76 |
"163": "NON_PARTICIPATING",
|
| 77 |
"164": "SUCCESS",
|
| 78 |
"165": "NON_PARTICIPATING",
|
|
@@ -84,17 +84,17 @@
|
|
| 84 |
"170": "NON_PARTICIPATING",
|
| 85 |
"171": "NON_PARTICIPATING",
|
| 86 |
"172": "NON_PARTICIPATING",
|
| 87 |
-
"173": "
|
| 88 |
"174": "NON_PARTICIPATING",
|
| 89 |
-
"175": "
|
| 90 |
"176": "NON_PARTICIPATING",
|
| 91 |
"177": "NON_PARTICIPATING",
|
| 92 |
"178": "NON_PARTICIPATING",
|
| 93 |
"179": "NON_PARTICIPATING",
|
| 94 |
"18": "SUCCESS",
|
| 95 |
-
"180": "
|
| 96 |
"181": "NON_PARTICIPATING",
|
| 97 |
-
"182": "
|
| 98 |
"183": "NON_PARTICIPATING",
|
| 99 |
"184": "NON_PARTICIPATING",
|
| 100 |
"185": "NON_PARTICIPATING",
|
|
@@ -110,7 +110,7 @@
|
|
| 110 |
"194": "NON_PARTICIPATING",
|
| 111 |
"195": "NON_PARTICIPATING",
|
| 112 |
"196": "NON_PARTICIPATING",
|
| 113 |
-
"197": "
|
| 114 |
"198": "NON_PARTICIPATING",
|
| 115 |
"199": "NON_PARTICIPATING",
|
| 116 |
"2": "SUCCESS",
|
|
@@ -118,15 +118,15 @@
|
|
| 118 |
"200": "NON_PARTICIPATING",
|
| 119 |
"201": "NON_PARTICIPATING",
|
| 120 |
"202": "NON_PARTICIPATING",
|
| 121 |
-
"203": "
|
| 122 |
"204": "NON_PARTICIPATING",
|
| 123 |
"205": "NON_PARTICIPATING",
|
| 124 |
"206": "NON_PARTICIPATING",
|
| 125 |
"207": "NON_PARTICIPATING",
|
| 126 |
"208": "NON_PARTICIPATING",
|
| 127 |
-
"209": "
|
| 128 |
"21": "SUCCESS",
|
| 129 |
-
"210": "
|
| 130 |
"211": "NON_PARTICIPATING",
|
| 131 |
"212": "NON_PARTICIPATING",
|
| 132 |
"213": "NON_PARTICIPATING",
|
|
@@ -150,8 +150,8 @@
|
|
| 150 |
"23": "SUCCESS",
|
| 151 |
"230": "NON_PARTICIPATING",
|
| 152 |
"231": "NON_PARTICIPATING",
|
| 153 |
-
"232": "
|
| 154 |
-
"233": "
|
| 155 |
"234": "NON_PARTICIPATING",
|
| 156 |
"235": "NON_PARTICIPATING",
|
| 157 |
"236": "NON_PARTICIPATING",
|
|
@@ -159,7 +159,7 @@
|
|
| 159 |
"238": "NON_PARTICIPATING",
|
| 160 |
"239": "NON_PARTICIPATING",
|
| 161 |
"24": "SUCCESS",
|
| 162 |
-
"240": "
|
| 163 |
"241": "NON_PARTICIPATING",
|
| 164 |
"242": "NON_PARTICIPATING",
|
| 165 |
"243": "NON_PARTICIPATING",
|
|
@@ -196,14 +196,14 @@
|
|
| 196 |
"41": "SUCCESS",
|
| 197 |
"42": "SUCCESS",
|
| 198 |
"43": "SUCCESS",
|
| 199 |
-
"44": "
|
| 200 |
"45": "SUCCESS",
|
| 201 |
"46": "SUCCESS",
|
| 202 |
"47": "SUCCESS",
|
| 203 |
"48": "NON_PARTICIPATING",
|
| 204 |
"49": "SUCCESS",
|
| 205 |
"5": "SUCCESS",
|
| 206 |
-
"50": "
|
| 207 |
"51": "SUCCESS",
|
| 208 |
"52": "NON_PARTICIPATING",
|
| 209 |
"53": "SUCCESS",
|
|
@@ -212,24 +212,24 @@
|
|
| 212 |
"56": "SUCCESS",
|
| 213 |
"57": "SUCCESS",
|
| 214 |
"58": "NON_PARTICIPATING",
|
| 215 |
-
"59": "
|
| 216 |
"6": "SUCCESS",
|
| 217 |
"60": "SUCCESS",
|
| 218 |
"61": "SUCCESS",
|
| 219 |
"62": "SUCCESS",
|
| 220 |
"63": "SUCCESS",
|
| 221 |
"64": "SUCCESS",
|
| 222 |
-
"65": "
|
| 223 |
"66": "NON_PARTICIPATING",
|
| 224 |
"67": "SUCCESS",
|
| 225 |
-
"68": "
|
| 226 |
"69": "SUCCESS",
|
| 227 |
"7": "SUCCESS",
|
| 228 |
"70": "SUCCESS",
|
| 229 |
"71": "SUCCESS",
|
| 230 |
"72": "SUCCESS",
|
| 231 |
"73": "SUCCESS",
|
| 232 |
-
"74": "
|
| 233 |
"75": "SUCCESS",
|
| 234 |
"76": "SUCCESS",
|
| 235 |
"77": "SUCCESS",
|
|
@@ -256,7 +256,7 @@
|
|
| 256 |
"96": "SUCCESS",
|
| 257 |
"97": "SUCCESS",
|
| 258 |
"98": "SUCCESS",
|
| 259 |
-
"99": "
|
| 260 |
},
|
| 261 |
"architectures": [
|
| 262 |
"LlamaForCausalLM"
|
|
@@ -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",
|
|
|
|
| 5 |
"1": "SUCCESS",
|
| 6 |
"10": "NON_PARTICIPATING",
|
| 7 |
"100": "NON_PARTICIPATING",
|
| 8 |
+
"101": "NON_PARTICIPATING",
|
| 9 |
"102": "NON_PARTICIPATING",
|
| 10 |
"103": "NON_PARTICIPATING",
|
| 11 |
"104": "NON_PARTICIPATING",
|
| 12 |
"105": "NON_PARTICIPATING",
|
| 13 |
"106": "NON_PARTICIPATING",
|
| 14 |
+
"107": "NON_PARTICIPATING",
|
| 15 |
"108": "NON_PARTICIPATING",
|
| 16 |
"109": "NON_PARTICIPATING",
|
| 17 |
"11": "SUCCESS",
|
| 18 |
"110": "NON_PARTICIPATING",
|
| 19 |
+
"111": "SUCCESS",
|
| 20 |
"112": "NON_PARTICIPATING",
|
| 21 |
"113": "NON_PARTICIPATING",
|
| 22 |
"114": "NON_PARTICIPATING",
|
| 23 |
"115": "NON_PARTICIPATING",
|
| 24 |
"116": "NON_PARTICIPATING",
|
| 25 |
+
"117": "NON_PARTICIPATING",
|
| 26 |
"118": "NON_PARTICIPATING",
|
| 27 |
"119": "NON_PARTICIPATING",
|
| 28 |
"12": "SUCCESS",
|
| 29 |
+
"120": "SUCCESS",
|
| 30 |
"121": "NON_PARTICIPATING",
|
| 31 |
"122": "NON_PARTICIPATING",
|
| 32 |
"123": "NON_PARTICIPATING",
|
|
|
|
| 40 |
"130": "NON_PARTICIPATING",
|
| 41 |
"131": "NON_PARTICIPATING",
|
| 42 |
"132": "NON_PARTICIPATING",
|
| 43 |
+
"133": "SUCCESS",
|
| 44 |
"134": "NON_PARTICIPATING",
|
| 45 |
"135": "NON_PARTICIPATING",
|
| 46 |
"136": "NON_PARTICIPATING",
|
|
|
|
| 53 |
"142": "NON_PARTICIPATING",
|
| 54 |
"143": "NON_PARTICIPATING",
|
| 55 |
"144": "NON_PARTICIPATING",
|
| 56 |
+
"145": "SUCCESS",
|
| 57 |
+
"146": "NON_PARTICIPATING",
|
| 58 |
"147": "NON_PARTICIPATING",
|
| 59 |
+
"148": "SUCCESS",
|
| 60 |
"149": "NON_PARTICIPATING",
|
| 61 |
"15": "SUCCESS",
|
| 62 |
"150": "NON_PARTICIPATING",
|
|
|
|
| 64 |
"152": "NON_PARTICIPATING",
|
| 65 |
"153": "NON_PARTICIPATING",
|
| 66 |
"154": "NON_PARTICIPATING",
|
| 67 |
+
"155": "SUCCESS",
|
| 68 |
"156": "NON_PARTICIPATING",
|
| 69 |
"157": "NON_PARTICIPATING",
|
| 70 |
"158": "NON_PARTICIPATING",
|
|
|
|
| 72 |
"16": "SUCCESS",
|
| 73 |
"160": "NON_PARTICIPATING",
|
| 74 |
"161": "NON_PARTICIPATING",
|
| 75 |
+
"162": "NON_PARTICIPATING",
|
| 76 |
"163": "NON_PARTICIPATING",
|
| 77 |
"164": "SUCCESS",
|
| 78 |
"165": "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": "NON_PARTICIPATING",
|
| 91 |
"177": "NON_PARTICIPATING",
|
| 92 |
"178": "NON_PARTICIPATING",
|
| 93 |
"179": "NON_PARTICIPATING",
|
| 94 |
"18": "SUCCESS",
|
| 95 |
+
"180": "NON_PARTICIPATING",
|
| 96 |
"181": "NON_PARTICIPATING",
|
| 97 |
+
"182": "SUCCESS",
|
| 98 |
"183": "NON_PARTICIPATING",
|
| 99 |
"184": "NON_PARTICIPATING",
|
| 100 |
"185": "NON_PARTICIPATING",
|
|
|
|
| 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": "SUCCESS",
|
|
|
|
| 118 |
"200": "NON_PARTICIPATING",
|
| 119 |
"201": "NON_PARTICIPATING",
|
| 120 |
"202": "NON_PARTICIPATING",
|
| 121 |
+
"203": "NON_PARTICIPATING",
|
| 122 |
"204": "NON_PARTICIPATING",
|
| 123 |
"205": "NON_PARTICIPATING",
|
| 124 |
"206": "NON_PARTICIPATING",
|
| 125 |
"207": "NON_PARTICIPATING",
|
| 126 |
"208": "NON_PARTICIPATING",
|
| 127 |
+
"209": "NON_PARTICIPATING",
|
| 128 |
"21": "SUCCESS",
|
| 129 |
+
"210": "NON_PARTICIPATING",
|
| 130 |
"211": "NON_PARTICIPATING",
|
| 131 |
"212": "NON_PARTICIPATING",
|
| 132 |
"213": "NON_PARTICIPATING",
|
|
|
|
| 150 |
"23": "SUCCESS",
|
| 151 |
"230": "NON_PARTICIPATING",
|
| 152 |
"231": "NON_PARTICIPATING",
|
| 153 |
+
"232": "SUCCESS",
|
| 154 |
+
"233": "SUCCESS",
|
| 155 |
"234": "NON_PARTICIPATING",
|
| 156 |
"235": "NON_PARTICIPATING",
|
| 157 |
"236": "NON_PARTICIPATING",
|
|
|
|
| 159 |
"238": "NON_PARTICIPATING",
|
| 160 |
"239": "NON_PARTICIPATING",
|
| 161 |
"24": "SUCCESS",
|
| 162 |
+
"240": "SUCCESS",
|
| 163 |
"241": "NON_PARTICIPATING",
|
| 164 |
"242": "NON_PARTICIPATING",
|
| 165 |
"243": "NON_PARTICIPATING",
|
|
|
|
| 196 |
"41": "SUCCESS",
|
| 197 |
"42": "SUCCESS",
|
| 198 |
"43": "SUCCESS",
|
| 199 |
+
"44": "NON_PARTICIPATING",
|
| 200 |
"45": "SUCCESS",
|
| 201 |
"46": "SUCCESS",
|
| 202 |
"47": "SUCCESS",
|
| 203 |
"48": "NON_PARTICIPATING",
|
| 204 |
"49": "SUCCESS",
|
| 205 |
"5": "SUCCESS",
|
| 206 |
+
"50": "NON_PARTICIPATING",
|
| 207 |
"51": "SUCCESS",
|
| 208 |
"52": "NON_PARTICIPATING",
|
| 209 |
"53": "SUCCESS",
|
|
|
|
| 212 |
"56": "SUCCESS",
|
| 213 |
"57": "SUCCESS",
|
| 214 |
"58": "NON_PARTICIPATING",
|
| 215 |
+
"59": "SUCCESS",
|
| 216 |
"6": "SUCCESS",
|
| 217 |
"60": "SUCCESS",
|
| 218 |
"61": "SUCCESS",
|
| 219 |
"62": "SUCCESS",
|
| 220 |
"63": "SUCCESS",
|
| 221 |
"64": "SUCCESS",
|
| 222 |
+
"65": "SUCCESS",
|
| 223 |
"66": "NON_PARTICIPATING",
|
| 224 |
"67": "SUCCESS",
|
| 225 |
+
"68": "NON_PARTICIPATING",
|
| 226 |
"69": "SUCCESS",
|
| 227 |
"7": "SUCCESS",
|
| 228 |
"70": "SUCCESS",
|
| 229 |
"71": "SUCCESS",
|
| 230 |
"72": "SUCCESS",
|
| 231 |
"73": "SUCCESS",
|
| 232 |
+
"74": "SUCCESS",
|
| 233 |
"75": "SUCCESS",
|
| 234 |
"76": "SUCCESS",
|
| 235 |
"77": "SUCCESS",
|
|
|
|
| 256 |
"96": "SUCCESS",
|
| 257 |
"97": "SUCCESS",
|
| 258 |
"98": "SUCCESS",
|
| 259 |
+
"99": "SUCCESS"
|
| 260 |
},
|
| 261 |
"architectures": [
|
| 262 |
"LlamaForCausalLM"
|
|
|
|
| 264 |
"attention_bias": false,
|
| 265 |
"attention_dropout": 0.0,
|
| 266 |
"block_list": [
|
| 267 |
+
5707764,
|
| 268 |
+
5707794
|
| 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": 2,
|
| 276 |
"intermediate_size": 5632,
|
| 277 |
+
"last_allreduce_block": 5707679,
|
| 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 8800564012
|
|
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:9b4ad41b6291b149848a0c0db9d162f1ee7aaf31d11ba68252b52723ab87732f
|
| 3 |
size 8800564012
|
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 sha256:0ac4ed29b35c63b2c3e95c0e4654d3a1a591f24c48bfb24cbc296b7612396763
|
| 3 |
size 4400216536
|