Run 5. Outer Step 54. Inner Step 2.
Browse files- config.json +17 -17
- inner_optimizer.pt +1 -1
- model.safetensors +1 -1
config.json
CHANGED
|
@@ -8,7 +8,7 @@
|
|
| 8 |
"101": "NON_PARTICIPATING",
|
| 9 |
"102": "NON_PARTICIPATING",
|
| 10 |
"103": "NON_PARTICIPATING",
|
| 11 |
-
"104": "
|
| 12 |
"105": "NON_PARTICIPATING",
|
| 13 |
"106": "NON_PARTICIPATING",
|
| 14 |
"107": "NON_PARTICIPATING",
|
|
@@ -16,7 +16,7 @@
|
|
| 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",
|
|
@@ -52,13 +52,13 @@
|
|
| 52 |
"141": "NON_PARTICIPATING",
|
| 53 |
"142": "NON_PARTICIPATING",
|
| 54 |
"143": "NON_PARTICIPATING",
|
| 55 |
-
"144": "
|
| 56 |
"145": "NON_PARTICIPATING",
|
| 57 |
"146": "NON_PARTICIPATING",
|
| 58 |
"147": "NON_PARTICIPATING",
|
| 59 |
"148": "NON_PARTICIPATING",
|
| 60 |
"149": "NON_PARTICIPATING",
|
| 61 |
-
"15": "
|
| 62 |
"150": "NON_PARTICIPATING",
|
| 63 |
"151": "NON_PARTICIPATING",
|
| 64 |
"152": "NON_PARTICIPATING",
|
|
@@ -74,7 +74,7 @@
|
|
| 74 |
"161": "NON_PARTICIPATING",
|
| 75 |
"162": "NON_PARTICIPATING",
|
| 76 |
"163": "NON_PARTICIPATING",
|
| 77 |
-
"164": "
|
| 78 |
"165": "NON_PARTICIPATING",
|
| 79 |
"166": "NON_PARTICIPATING",
|
| 80 |
"167": "NON_PARTICIPATING",
|
|
@@ -84,12 +84,12 @@
|
|
| 84 |
"170": "NON_PARTICIPATING",
|
| 85 |
"171": "NON_PARTICIPATING",
|
| 86 |
"172": "NON_PARTICIPATING",
|
| 87 |
-
"173": "
|
| 88 |
"174": "NON_PARTICIPATING",
|
| 89 |
"175": "NON_PARTICIPATING",
|
| 90 |
"176": "NON_PARTICIPATING",
|
| 91 |
"177": "NON_PARTICIPATING",
|
| 92 |
-
"178": "
|
| 93 |
"179": "NON_PARTICIPATING",
|
| 94 |
"18": "SUCCESS",
|
| 95 |
"180": "NON_PARTICIPATING",
|
|
@@ -126,8 +126,8 @@
|
|
| 126 |
"208": "NON_PARTICIPATING",
|
| 127 |
"209": "NON_PARTICIPATING",
|
| 128 |
"21": "SUCCESS",
|
| 129 |
-
"210": "
|
| 130 |
-
"211": "
|
| 131 |
"212": "NON_PARTICIPATING",
|
| 132 |
"213": "NON_PARTICIPATING",
|
| 133 |
"214": "NON_PARTICIPATING",
|
|
@@ -159,7 +159,7 @@
|
|
| 159 |
"238": "NON_PARTICIPATING",
|
| 160 |
"239": "NON_PARTICIPATING",
|
| 161 |
"24": "NON_PARTICIPATING",
|
| 162 |
-
"240": "
|
| 163 |
"241": "NON_PARTICIPATING",
|
| 164 |
"242": "NON_PARTICIPATING",
|
| 165 |
"243": "NON_PARTICIPATING",
|
|
@@ -183,7 +183,7 @@
|
|
| 183 |
"3": "SUCCESS",
|
| 184 |
"30": "NON_PARTICIPATING",
|
| 185 |
"31": "NON_PARTICIPATING",
|
| 186 |
-
"32": "
|
| 187 |
"33": "NON_PARTICIPATING",
|
| 188 |
"34": "NON_PARTICIPATING",
|
| 189 |
"35": "NON_PARTICIPATING",
|
|
@@ -218,7 +218,7 @@
|
|
| 218 |
"61": "NON_PARTICIPATING",
|
| 219 |
"62": "SUCCESS",
|
| 220 |
"63": "NON_PARTICIPATING",
|
| 221 |
-
"64": "
|
| 222 |
"65": "SUCCESS",
|
| 223 |
"66": "NON_PARTICIPATING",
|
| 224 |
"67": "SUCCESS",
|
|
@@ -252,7 +252,7 @@
|
|
| 252 |
"92": "SUCCESS",
|
| 253 |
"93": "SUCCESS",
|
| 254 |
"94": "NON_PARTICIPATING",
|
| 255 |
-
"95": "
|
| 256 |
"96": "NON_PARTICIPATING",
|
| 257 |
"97": "NON_PARTICIPATING",
|
| 258 |
"98": "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",
|
|
|
|
| 8 |
"101": "NON_PARTICIPATING",
|
| 9 |
"102": "NON_PARTICIPATING",
|
| 10 |
"103": "NON_PARTICIPATING",
|
| 11 |
+
"104": "SUCCESS",
|
| 12 |
"105": "NON_PARTICIPATING",
|
| 13 |
"106": "NON_PARTICIPATING",
|
| 14 |
"107": "NON_PARTICIPATING",
|
|
|
|
| 16 |
"109": "NON_PARTICIPATING",
|
| 17 |
"11": "SUCCESS",
|
| 18 |
"110": "NON_PARTICIPATING",
|
| 19 |
+
"111": "NON_PARTICIPATING",
|
| 20 |
"112": "NON_PARTICIPATING",
|
| 21 |
"113": "NON_PARTICIPATING",
|
| 22 |
"114": "NON_PARTICIPATING",
|
|
|
|
| 52 |
"141": "NON_PARTICIPATING",
|
| 53 |
"142": "NON_PARTICIPATING",
|
| 54 |
"143": "NON_PARTICIPATING",
|
| 55 |
+
"144": "NON_PARTICIPATING",
|
| 56 |
"145": "NON_PARTICIPATING",
|
| 57 |
"146": "NON_PARTICIPATING",
|
| 58 |
"147": "NON_PARTICIPATING",
|
| 59 |
"148": "NON_PARTICIPATING",
|
| 60 |
"149": "NON_PARTICIPATING",
|
| 61 |
+
"15": "SUCCESS",
|
| 62 |
"150": "NON_PARTICIPATING",
|
| 63 |
"151": "NON_PARTICIPATING",
|
| 64 |
"152": "NON_PARTICIPATING",
|
|
|
|
| 74 |
"161": "NON_PARTICIPATING",
|
| 75 |
"162": "NON_PARTICIPATING",
|
| 76 |
"163": "NON_PARTICIPATING",
|
| 77 |
+
"164": "SUCCESS",
|
| 78 |
"165": "NON_PARTICIPATING",
|
| 79 |
"166": "NON_PARTICIPATING",
|
| 80 |
"167": "NON_PARTICIPATING",
|
|
|
|
| 84 |
"170": "NON_PARTICIPATING",
|
| 85 |
"171": "NON_PARTICIPATING",
|
| 86 |
"172": "NON_PARTICIPATING",
|
| 87 |
+
"173": "SUCCESS",
|
| 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",
|
|
|
|
| 126 |
"208": "NON_PARTICIPATING",
|
| 127 |
"209": "NON_PARTICIPATING",
|
| 128 |
"21": "SUCCESS",
|
| 129 |
+
"210": "SUCCESS",
|
| 130 |
+
"211": "SUCCESS",
|
| 131 |
"212": "NON_PARTICIPATING",
|
| 132 |
"213": "NON_PARTICIPATING",
|
| 133 |
"214": "NON_PARTICIPATING",
|
|
|
|
| 159 |
"238": "NON_PARTICIPATING",
|
| 160 |
"239": "NON_PARTICIPATING",
|
| 161 |
"24": "NON_PARTICIPATING",
|
| 162 |
+
"240": "SUCCESS",
|
| 163 |
"241": "NON_PARTICIPATING",
|
| 164 |
"242": "NON_PARTICIPATING",
|
| 165 |
"243": "NON_PARTICIPATING",
|
|
|
|
| 183 |
"3": "SUCCESS",
|
| 184 |
"30": "NON_PARTICIPATING",
|
| 185 |
"31": "NON_PARTICIPATING",
|
| 186 |
+
"32": "SUCCESS",
|
| 187 |
"33": "NON_PARTICIPATING",
|
| 188 |
"34": "NON_PARTICIPATING",
|
| 189 |
"35": "NON_PARTICIPATING",
|
|
|
|
| 218 |
"61": "NON_PARTICIPATING",
|
| 219 |
"62": "SUCCESS",
|
| 220 |
"63": "NON_PARTICIPATING",
|
| 221 |
+
"64": "SUCCESS",
|
| 222 |
"65": "SUCCESS",
|
| 223 |
"66": "NON_PARTICIPATING",
|
| 224 |
"67": "SUCCESS",
|
|
|
|
| 252 |
"92": "SUCCESS",
|
| 253 |
"93": "SUCCESS",
|
| 254 |
"94": "NON_PARTICIPATING",
|
| 255 |
+
"95": "SUCCESS",
|
| 256 |
"96": "NON_PARTICIPATING",
|
| 257 |
"97": "NON_PARTICIPATING",
|
| 258 |
"98": "SUCCESS",
|
|
|
|
| 264 |
"attention_bias": false,
|
| 265 |
"attention_dropout": 0.0,
|
| 266 |
"block_list": [
|
| 267 |
+
5778220,
|
| 268 |
+
5778249
|
| 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": 5778130,
|
| 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 8800564076
|
|
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:56b058db55bb7d2f1f9370f3c90efddaa7808c9425792074ebcacf3f1491cdd2
|
| 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 sha256:0325f421eb74d2be905cf68f1ff7c76371d0abc198050e8ef30919f84327ada5
|
| 3 |
size 4400216536
|