mizzaay commited on
Commit
af5326d
·
verified ·
1 Parent(s): f37821a

Run 6. Outer Step 44. Inner Step 5.

Browse files
Files changed (4) hide show
  1. config.json +17 -17
  2. gradients.pt +1 -1
  3. inner_optimizer.pt +1 -1
  4. 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": "SUCCESS",
22
  "114": "NOT_ALIVE",
23
  "115": "NOT_ALIVE",
24
  "116": "NOT_ALIVE",
@@ -27,7 +27,7 @@
27
  "119": "NOT_ALIVE",
28
  "12": "NOT_ALIVE",
29
  "120": "NOT_ALIVE",
30
- "121": "NOT_ALIVE",
31
  "122": "NOT_ALIVE",
32
  "123": "NOT_ALIVE",
33
  "124": "NON_PARTICIPATING",
@@ -54,7 +54,7 @@
54
  "143": "NOT_ALIVE",
55
  "144": "NOT_ALIVE",
56
  "145": "NOT_ALIVE",
57
- "146": "NOT_ALIVE",
58
  "147": "NOT_ALIVE",
59
  "148": "NOT_ALIVE",
60
  "149": "NOT_ALIVE",
@@ -101,7 +101,7 @@
101
  "186": "NOT_ALIVE",
102
  "187": "SUCCESS",
103
  "188": "NOT_ALIVE",
104
- "189": "NOT_ALIVE",
105
  "19": "NOT_ALIVE",
106
  "190": "SUCCESS",
107
  "191": "NON_PARTICIPATING",
@@ -112,7 +112,7 @@
112
  "196": "NOT_ALIVE",
113
  "197": "NOT_ALIVE",
114
  "198": "NOT_ALIVE",
115
- "199": "SUCCESS",
116
  "2": "NOT_ALIVE",
117
  "20": "NOT_ALIVE",
118
  "200": "NOT_ALIVE",
@@ -130,7 +130,7 @@
130
  "211": "NOT_ALIVE",
131
  "212": "SUCCESS",
132
  "213": "NOT_ALIVE",
133
- "214": "NOT_ALIVE",
134
  "215": "NON_PARTICIPATING",
135
  "216": "NOT_ALIVE",
136
  "217": "SUCCESS",
@@ -141,7 +141,7 @@
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",
@@ -163,7 +163,7 @@
163
  "241": "NOT_ALIVE",
164
  "242": "SUCCESS",
165
  "243": "NOT_ALIVE",
166
- "244": "NON_PARTICIPATING",
167
  "245": "NOT_ALIVE",
168
  "246": "NOT_ALIVE",
169
  "247": "NOT_ALIVE",
@@ -182,7 +182,7 @@
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",
@@ -209,7 +209,7 @@
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",
@@ -226,7 +226,7 @@
226
  "69": "NOT_ALIVE",
227
  "7": "NOT_ALIVE",
228
  "70": "SUCCESS",
229
- "71": "FAIL",
230
  "72": "NOT_ALIVE",
231
  "73": "NOT_ALIVE",
232
  "74": "NOT_ALIVE",
@@ -239,7 +239,7 @@
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",
@@ -253,7 +253,7 @@
253
  "93": "NOT_ALIVE",
254
  "94": "NOT_ALIVE",
255
  "95": "NON_PARTICIPATING",
256
- "96": "NON_PARTICIPATING",
257
  "97": "NOT_ALIVE",
258
  "98": "NOT_ALIVE",
259
  "99": "SUCCESS"
@@ -264,17 +264,17 @@
264
  "attention_bias": false,
265
  "attention_dropout": 0.0,
266
  "block_list": [
267
- 6158285,
268
- 6158307
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": 91,
276
  "intermediate_size": 5632,
277
- "last_allreduce_block": 6157557,
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": "NOT_ALIVE",
22
  "114": "NOT_ALIVE",
23
  "115": "NOT_ALIVE",
24
  "116": "NOT_ALIVE",
 
27
  "119": "NOT_ALIVE",
28
  "12": "NOT_ALIVE",
29
  "120": "NOT_ALIVE",
30
+ "121": "SUCCESS",
31
  "122": "NOT_ALIVE",
32
  "123": "NOT_ALIVE",
33
  "124": "NON_PARTICIPATING",
 
54
  "143": "NOT_ALIVE",
55
  "144": "NOT_ALIVE",
56
  "145": "NOT_ALIVE",
57
+ "146": "NON_PARTICIPATING",
58
  "147": "NOT_ALIVE",
59
  "148": "NOT_ALIVE",
60
  "149": "NOT_ALIVE",
 
101
  "186": "NOT_ALIVE",
102
  "187": "SUCCESS",
103
  "188": "NOT_ALIVE",
104
+ "189": "SUCCESS",
105
  "19": "NOT_ALIVE",
106
  "190": "SUCCESS",
107
  "191": "NON_PARTICIPATING",
 
112
  "196": "NOT_ALIVE",
113
  "197": "NOT_ALIVE",
114
  "198": "NOT_ALIVE",
115
+ "199": "NOT_ALIVE",
116
  "2": "NOT_ALIVE",
117
  "20": "NOT_ALIVE",
118
  "200": "NOT_ALIVE",
 
130
  "211": "NOT_ALIVE",
131
  "212": "SUCCESS",
132
  "213": "NOT_ALIVE",
133
+ "214": "NON_PARTICIPATING",
134
  "215": "NON_PARTICIPATING",
135
  "216": "NOT_ALIVE",
136
  "217": "SUCCESS",
 
141
  "221": "NOT_ALIVE",
142
  "222": "NOT_ALIVE",
143
  "223": "NOT_ALIVE",
144
+ "224": "NOT_ALIVE",
145
  "225": "NOT_ALIVE",
146
  "226": "NON_PARTICIPATING",
147
  "227": "NOT_ALIVE",
 
163
  "241": "NOT_ALIVE",
164
  "242": "SUCCESS",
165
  "243": "NOT_ALIVE",
166
+ "244": "SUCCESS",
167
  "245": "NOT_ALIVE",
168
  "246": "NOT_ALIVE",
169
  "247": "NOT_ALIVE",
 
182
  "29": "NOT_ALIVE",
183
  "3": "SUCCESS",
184
  "30": "NOT_ALIVE",
185
+ "31": "SUCCESS",
186
  "32": "SUCCESS",
187
  "33": "NOT_ALIVE",
188
  "34": "NOT_ALIVE",
 
209
  "53": "SUCCESS",
210
  "54": "NOT_ALIVE",
211
  "55": "NOT_ALIVE",
212
+ "56": "NON_PARTICIPATING",
213
  "57": "NON_PARTICIPATING",
214
  "58": "NOT_ALIVE",
215
  "59": "NOT_ALIVE",
 
226
  "69": "NOT_ALIVE",
227
  "7": "NOT_ALIVE",
228
  "70": "SUCCESS",
229
+ "71": "NON_PARTICIPATING",
230
  "72": "NOT_ALIVE",
231
  "73": "NOT_ALIVE",
232
  "74": "NOT_ALIVE",
 
239
  "80": "NOT_ALIVE",
240
  "81": "NOT_ALIVE",
241
  "82": "NOT_ALIVE",
242
+ "83": "NOT_ALIVE",
243
  "84": "NOT_ALIVE",
244
  "85": "NOT_ALIVE",
245
  "86": "NOT_ALIVE",
 
253
  "93": "NOT_ALIVE",
254
  "94": "NOT_ALIVE",
255
  "95": "NON_PARTICIPATING",
256
+ "96": "NOT_ALIVE",
257
  "97": "NOT_ALIVE",
258
  "98": "NOT_ALIVE",
259
  "99": "SUCCESS"
 
264
  "attention_bias": false,
265
  "attention_dropout": 0.0,
266
  "block_list": [
267
+ 6158423,
268
+ 6158439
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": 6158311,
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 sha256:58cff972cbb29610341ff230818445fd6adab2d4a3ae0db67a29d1d90b947b1c
3
  size 88389626
 
1
  version https://git-lfs.github.com/spec/v1
2
+ oid sha256:819242b9e7b6feb9c25b111880b2ad8c75a1ba23d3b4f8a4000fd087612a2daf
3
  size 88389626
inner_optimizer.pt CHANGED
@@ -1,3 +1,3 @@
1
  version https://git-lfs.github.com/spec/v1
2
- oid sha256:fa8b202555a707f1545e8cec1ea7019b703377082aaaff17b5acba94c3cd4e1a
3
  size 8800564076
 
1
  version https://git-lfs.github.com/spec/v1
2
+ oid sha256:e2811d7b4598b28eb678b2f3987a3a6a36db06728803be73236ba7a65db89489
3
  size 8800564076
model.safetensors CHANGED
@@ -1,3 +1,3 @@
1
  version https://git-lfs.github.com/spec/v1
2
- oid sha256:3a8b8554b55dfa327971a58903a3e46d0602a3e8e0dbcc4f8abaa3697fad8a2f
3
  size 4400216536
 
1
  version https://git-lfs.github.com/spec/v1
2
+ oid sha256:c352ba3b32a0d825ae76c68dd1df17dbb85d4fd8b2e26f8015eebb3e361e8cc0
3
  size 4400216536