Run 5. Outer Step 94. Inner Step 2.
Browse files- config.json +37 -37
- inner_optimizer.pt +1 -1
- model.safetensors +1 -1
config.json
CHANGED
@@ -24,37 +24,37 @@
|
|
24 |
"116": "SUCCESS",
|
25 |
"117": "NON_PARTICIPATING",
|
26 |
"118": "NON_PARTICIPATING",
|
27 |
-
"119": "
|
28 |
"12": "NON_PARTICIPATING",
|
29 |
"120": "NOT_ALIVE",
|
30 |
"121": "NOT_ALIVE",
|
31 |
"122": "NOT_ALIVE",
|
32 |
"123": "NOT_ALIVE",
|
33 |
-
"124": "
|
34 |
"125": "NON_PARTICIPATING",
|
35 |
"126": "NON_PARTICIPATING",
|
36 |
"127": "NOT_ALIVE",
|
37 |
"128": "NOT_ALIVE",
|
38 |
-
"129": "
|
39 |
"13": "NOT_ALIVE",
|
40 |
-
"130": "
|
41 |
"131": "NOT_ALIVE",
|
42 |
-
"132": "
|
43 |
"133": "SUCCESS",
|
44 |
-
"134": "
|
45 |
"135": "NON_PARTICIPATING",
|
46 |
"136": "NOT_ALIVE",
|
47 |
-
"137": "
|
48 |
"138": "NON_PARTICIPATING",
|
49 |
"139": "NON_PARTICIPATING",
|
50 |
"14": "NON_PARTICIPATING",
|
51 |
"140": "NOT_ALIVE",
|
52 |
"141": "SUCCESS",
|
53 |
-
"142": "
|
54 |
"143": "NON_PARTICIPATING",
|
55 |
"144": "NOT_ALIVE",
|
56 |
"145": "NOT_ALIVE",
|
57 |
-
"146": "
|
58 |
"147": "SUCCESS",
|
59 |
"148": "SUCCESS",
|
60 |
"149": "NOT_ALIVE",
|
@@ -69,8 +69,8 @@
|
|
69 |
"157": "NOT_ALIVE",
|
70 |
"158": "NON_PARTICIPATING",
|
71 |
"159": "NOT_ALIVE",
|
72 |
-
"16": "
|
73 |
-
"160": "
|
74 |
"161": "NOT_ALIVE",
|
75 |
"162": "NON_PARTICIPATING",
|
76 |
"163": "NOT_ALIVE",
|
@@ -86,10 +86,10 @@
|
|
86 |
"172": "NOT_ALIVE",
|
87 |
"173": "NOT_ALIVE",
|
88 |
"174": "SUCCESS",
|
89 |
-
"175": "
|
90 |
"176": "NOT_ALIVE",
|
91 |
"177": "NON_PARTICIPATING",
|
92 |
-
"178": "
|
93 |
"179": "NOT_ALIVE",
|
94 |
"18": "SUCCESS",
|
95 |
"180": "NOT_ALIVE",
|
@@ -127,7 +127,7 @@
|
|
127 |
"209": "NOT_ALIVE",
|
128 |
"21": "NOT_ALIVE",
|
129 |
"210": "SUCCESS",
|
130 |
-
"211": "
|
131 |
"212": "NOT_ALIVE",
|
132 |
"213": "NON_PARTICIPATING",
|
133 |
"214": "NOT_ALIVE",
|
@@ -138,7 +138,7 @@
|
|
138 |
"219": "NON_PARTICIPATING",
|
139 |
"22": "SUCCESS",
|
140 |
"220": "SUCCESS",
|
141 |
-
"221": "
|
142 |
"222": "NOT_ALIVE",
|
143 |
"223": "NOT_ALIVE",
|
144 |
"224": "NOT_ALIVE",
|
@@ -168,21 +168,21 @@
|
|
168 |
"246": "NOT_ALIVE",
|
169 |
"247": "NON_PARTICIPATING",
|
170 |
"248": "NOT_ALIVE",
|
171 |
-
"249": "
|
172 |
"25": "SUCCESS",
|
173 |
-
"250": "
|
174 |
"251": "NOT_ALIVE",
|
175 |
"252": "NOT_ALIVE",
|
176 |
"253": "NOT_ALIVE",
|
177 |
"254": "NOT_ALIVE",
|
178 |
-
"255": "
|
179 |
-
"26": "
|
180 |
"27": "NON_PARTICIPATING",
|
181 |
"28": "SUCCESS",
|
182 |
"29": "NOT_ALIVE",
|
183 |
-
"3": "
|
184 |
"30": "NOT_ALIVE",
|
185 |
-
"31": "
|
186 |
"32": "NOT_ALIVE",
|
187 |
"33": "NOT_ALIVE",
|
188 |
"34": "NOT_ALIVE",
|
@@ -202,29 +202,29 @@
|
|
202 |
"47": "NON_PARTICIPATING",
|
203 |
"48": "SUCCESS",
|
204 |
"49": "NOT_ALIVE",
|
205 |
-
"5": "
|
206 |
"50": "NOT_ALIVE",
|
207 |
"51": "NOT_ALIVE",
|
208 |
"52": "NOT_ALIVE",
|
209 |
"53": "NON_PARTICIPATING",
|
210 |
-
"54": "
|
211 |
-
"55": "
|
212 |
"56": "SUCCESS",
|
213 |
"57": "SUCCESS",
|
214 |
"58": "SUCCESS",
|
215 |
"59": "NOT_ALIVE",
|
216 |
"6": "NOT_ALIVE",
|
217 |
-
"60": "
|
218 |
"61": "NOT_ALIVE",
|
219 |
"62": "NON_PARTICIPATING",
|
220 |
"63": "NON_PARTICIPATING",
|
221 |
"64": "NON_PARTICIPATING",
|
222 |
"65": "NOT_ALIVE",
|
223 |
-
"66": "
|
224 |
"67": "NOT_ALIVE",
|
225 |
"68": "SUCCESS",
|
226 |
"69": "SUCCESS",
|
227 |
-
"7": "
|
228 |
"70": "NOT_ALIVE",
|
229 |
"71": "SUCCESS",
|
230 |
"72": "NON_PARTICIPATING",
|
@@ -234,28 +234,28 @@
|
|
234 |
"76": "NON_PARTICIPATING",
|
235 |
"77": "NOT_ALIVE",
|
236 |
"78": "NOT_ALIVE",
|
237 |
-
"79": "
|
238 |
"8": "NOT_ALIVE",
|
239 |
"80": "NOT_ALIVE",
|
240 |
"81": "NOT_ALIVE",
|
241 |
"82": "NOT_ALIVE",
|
242 |
"83": "NOT_ALIVE",
|
243 |
"84": "SUCCESS",
|
244 |
-
"85": "
|
245 |
-
"86": "
|
246 |
"87": "SUCCESS",
|
247 |
"88": "NOT_ALIVE",
|
248 |
"89": "NOT_ALIVE",
|
249 |
"9": "NOT_ALIVE",
|
250 |
"90": "SUCCESS",
|
251 |
"91": "NOT_ALIVE",
|
252 |
-
"92": "
|
253 |
"93": "NON_PARTICIPATING",
|
254 |
"94": "NON_PARTICIPATING",
|
255 |
"95": "NOT_ALIVE",
|
256 |
"96": "NON_PARTICIPATING",
|
257 |
-
"97": "
|
258 |
-
"98": "
|
259 |
"99": "NOT_ALIVE"
|
260 |
},
|
261 |
"architectures": [
|
@@ -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",
|
|
|
24 |
"116": "SUCCESS",
|
25 |
"117": "NON_PARTICIPATING",
|
26 |
"118": "NON_PARTICIPATING",
|
27 |
+
"119": "SUCCESS",
|
28 |
"12": "NON_PARTICIPATING",
|
29 |
"120": "NOT_ALIVE",
|
30 |
"121": "NOT_ALIVE",
|
31 |
"122": "NOT_ALIVE",
|
32 |
"123": "NOT_ALIVE",
|
33 |
+
"124": "SUCCESS",
|
34 |
"125": "NON_PARTICIPATING",
|
35 |
"126": "NON_PARTICIPATING",
|
36 |
"127": "NOT_ALIVE",
|
37 |
"128": "NOT_ALIVE",
|
38 |
+
"129": "NON_PARTICIPATING",
|
39 |
"13": "NOT_ALIVE",
|
40 |
+
"130": "NOT_ALIVE",
|
41 |
"131": "NOT_ALIVE",
|
42 |
+
"132": "NON_PARTICIPATING",
|
43 |
"133": "SUCCESS",
|
44 |
+
"134": "NON_PARTICIPATING",
|
45 |
"135": "NON_PARTICIPATING",
|
46 |
"136": "NOT_ALIVE",
|
47 |
+
"137": "NOT_ALIVE",
|
48 |
"138": "NON_PARTICIPATING",
|
49 |
"139": "NON_PARTICIPATING",
|
50 |
"14": "NON_PARTICIPATING",
|
51 |
"140": "NOT_ALIVE",
|
52 |
"141": "SUCCESS",
|
53 |
+
"142": "SUCCESS",
|
54 |
"143": "NON_PARTICIPATING",
|
55 |
"144": "NOT_ALIVE",
|
56 |
"145": "NOT_ALIVE",
|
57 |
+
"146": "NON_PARTICIPATING",
|
58 |
"147": "SUCCESS",
|
59 |
"148": "SUCCESS",
|
60 |
"149": "NOT_ALIVE",
|
|
|
69 |
"157": "NOT_ALIVE",
|
70 |
"158": "NON_PARTICIPATING",
|
71 |
"159": "NOT_ALIVE",
|
72 |
+
"16": "NON_PARTICIPATING",
|
73 |
+
"160": "NOT_ALIVE",
|
74 |
"161": "NOT_ALIVE",
|
75 |
"162": "NON_PARTICIPATING",
|
76 |
"163": "NOT_ALIVE",
|
|
|
86 |
"172": "NOT_ALIVE",
|
87 |
"173": "NOT_ALIVE",
|
88 |
"174": "SUCCESS",
|
89 |
+
"175": "NON_PARTICIPATING",
|
90 |
"176": "NOT_ALIVE",
|
91 |
"177": "NON_PARTICIPATING",
|
92 |
+
"178": "NOT_ALIVE",
|
93 |
"179": "NOT_ALIVE",
|
94 |
"18": "SUCCESS",
|
95 |
"180": "NOT_ALIVE",
|
|
|
127 |
"209": "NOT_ALIVE",
|
128 |
"21": "NOT_ALIVE",
|
129 |
"210": "SUCCESS",
|
130 |
+
"211": "NON_PARTICIPATING",
|
131 |
"212": "NOT_ALIVE",
|
132 |
"213": "NON_PARTICIPATING",
|
133 |
"214": "NOT_ALIVE",
|
|
|
138 |
"219": "NON_PARTICIPATING",
|
139 |
"22": "SUCCESS",
|
140 |
"220": "SUCCESS",
|
141 |
+
"221": "NON_PARTICIPATING",
|
142 |
"222": "NOT_ALIVE",
|
143 |
"223": "NOT_ALIVE",
|
144 |
"224": "NOT_ALIVE",
|
|
|
168 |
"246": "NOT_ALIVE",
|
169 |
"247": "NON_PARTICIPATING",
|
170 |
"248": "NOT_ALIVE",
|
171 |
+
"249": "NOT_ALIVE",
|
172 |
"25": "SUCCESS",
|
173 |
+
"250": "NON_PARTICIPATING",
|
174 |
"251": "NOT_ALIVE",
|
175 |
"252": "NOT_ALIVE",
|
176 |
"253": "NOT_ALIVE",
|
177 |
"254": "NOT_ALIVE",
|
178 |
+
"255": "NON_PARTICIPATING",
|
179 |
+
"26": "NOT_ALIVE",
|
180 |
"27": "NON_PARTICIPATING",
|
181 |
"28": "SUCCESS",
|
182 |
"29": "NOT_ALIVE",
|
183 |
+
"3": "NOT_ALIVE",
|
184 |
"30": "NOT_ALIVE",
|
185 |
+
"31": "NOT_ALIVE",
|
186 |
"32": "NOT_ALIVE",
|
187 |
"33": "NOT_ALIVE",
|
188 |
"34": "NOT_ALIVE",
|
|
|
202 |
"47": "NON_PARTICIPATING",
|
203 |
"48": "SUCCESS",
|
204 |
"49": "NOT_ALIVE",
|
205 |
+
"5": "NON_PARTICIPATING",
|
206 |
"50": "NOT_ALIVE",
|
207 |
"51": "NOT_ALIVE",
|
208 |
"52": "NOT_ALIVE",
|
209 |
"53": "NON_PARTICIPATING",
|
210 |
+
"54": "NOT_ALIVE",
|
211 |
+
"55": "NOT_ALIVE",
|
212 |
"56": "SUCCESS",
|
213 |
"57": "SUCCESS",
|
214 |
"58": "SUCCESS",
|
215 |
"59": "NOT_ALIVE",
|
216 |
"6": "NOT_ALIVE",
|
217 |
+
"60": "SUCCESS",
|
218 |
"61": "NOT_ALIVE",
|
219 |
"62": "NON_PARTICIPATING",
|
220 |
"63": "NON_PARTICIPATING",
|
221 |
"64": "NON_PARTICIPATING",
|
222 |
"65": "NOT_ALIVE",
|
223 |
+
"66": "NOT_ALIVE",
|
224 |
"67": "NOT_ALIVE",
|
225 |
"68": "SUCCESS",
|
226 |
"69": "SUCCESS",
|
227 |
+
"7": "NOT_ALIVE",
|
228 |
"70": "NOT_ALIVE",
|
229 |
"71": "SUCCESS",
|
230 |
"72": "NON_PARTICIPATING",
|
|
|
234 |
"76": "NON_PARTICIPATING",
|
235 |
"77": "NOT_ALIVE",
|
236 |
"78": "NOT_ALIVE",
|
237 |
+
"79": "NOT_ALIVE",
|
238 |
"8": "NOT_ALIVE",
|
239 |
"80": "NOT_ALIVE",
|
240 |
"81": "NOT_ALIVE",
|
241 |
"82": "NOT_ALIVE",
|
242 |
"83": "NOT_ALIVE",
|
243 |
"84": "SUCCESS",
|
244 |
+
"85": "SUCCESS",
|
245 |
+
"86": "NON_PARTICIPATING",
|
246 |
"87": "SUCCESS",
|
247 |
"88": "NOT_ALIVE",
|
248 |
"89": "NOT_ALIVE",
|
249 |
"9": "NOT_ALIVE",
|
250 |
"90": "SUCCESS",
|
251 |
"91": "NOT_ALIVE",
|
252 |
+
"92": "SUCCESS",
|
253 |
"93": "NON_PARTICIPATING",
|
254 |
"94": "NON_PARTICIPATING",
|
255 |
"95": "NOT_ALIVE",
|
256 |
"96": "NON_PARTICIPATING",
|
257 |
+
"97": "SUCCESS",
|
258 |
+
"98": "NOT_ALIVE",
|
259 |
"99": "NOT_ALIVE"
|
260 |
},
|
261 |
"architectures": [
|
|
|
264 |
"attention_bias": false,
|
265 |
"attention_dropout": 0.0,
|
266 |
"block_list": [
|
267 |
+
5848823,
|
268 |
+
5848851
|
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": 5848758,
|
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 sha256:56319f166124601a58defde668989543d74df58ca6485f3095d2652916ce464b
|
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:d73850e6aca123aa4c6e7317bc474e1c06f391bd3b1258c82ca71e9c930e05f7
|
3 |
size 4400216536
|