Ham2363 commited on
Commit
b54ddb5
·
verified ·
1 Parent(s): 07d7b52

Run 5. Outer Step 151. Inner Step 2.

Browse files
Files changed (4) hide show
  1. config.json +107 -107
  2. gradients.pt +1 -1
  3. inner_optimizer.pt +1 -1
  4. model.safetensors +1 -1
config.json CHANGED
@@ -6,74 +6,74 @@
6
  "10": "NON_PARTICIPATING",
7
  "100": "NOT_ALIVE",
8
  "101": "NOT_ALIVE",
9
- "102": "NON_PARTICIPATING",
10
- "103": "NON_PARTICIPATING",
11
  "104": "NOT_ALIVE",
12
  "105": "NON_PARTICIPATING",
13
  "106": "NON_PARTICIPATING",
14
- "107": "NON_PARTICIPATING",
15
- "108": "NOT_ALIVE",
16
  "109": "NON_PARTICIPATING",
17
  "11": "NON_PARTICIPATING",
18
  "110": "NON_PARTICIPATING",
19
  "111": "NOT_ALIVE",
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": "NOT_ALIVE",
27
  "119": "NOT_ALIVE",
28
  "12": "NOT_ALIVE",
29
- "120": "NON_PARTICIPATING",
30
- "121": "NON_PARTICIPATING",
31
  "122": "NON_PARTICIPATING",
32
  "123": "NON_PARTICIPATING",
33
- "124": "NON_PARTICIPATING",
34
  "125": "NON_PARTICIPATING",
35
  "126": "NON_PARTICIPATING",
36
  "127": "NOT_ALIVE",
37
- "128": "NON_PARTICIPATING",
38
  "129": "NON_PARTICIPATING",
39
- "13": "NOT_ALIVE",
40
  "130": "NON_PARTICIPATING",
41
  "131": "NOT_ALIVE",
42
- "132": "NON_PARTICIPATING",
43
  "133": "NON_PARTICIPATING",
44
  "134": "NON_PARTICIPATING",
45
  "135": "NON_PARTICIPATING",
46
  "136": "NON_PARTICIPATING",
47
  "137": "NON_PARTICIPATING",
48
- "138": "NON_PARTICIPATING",
49
- "139": "NON_PARTICIPATING",
50
  "14": "NOT_ALIVE",
51
- "140": "NOT_ALIVE",
52
  "141": "NON_PARTICIPATING",
53
  "142": "NOT_ALIVE",
54
- "143": "NON_PARTICIPATING",
55
  "144": "NON_PARTICIPATING",
56
  "145": "NOT_ALIVE",
57
- "146": "NOT_ALIVE",
58
  "147": "NON_PARTICIPATING",
59
  "148": "NON_PARTICIPATING",
60
- "149": "NOT_ALIVE",
61
  "15": "NOT_ALIVE",
62
  "150": "NON_PARTICIPATING",
63
  "151": "NOT_ALIVE",
64
- "152": "NON_PARTICIPATING",
65
  "153": "NOT_ALIVE",
66
  "154": "NON_PARTICIPATING",
67
- "155": "NON_PARTICIPATING",
68
  "156": "NOT_ALIVE",
69
- "157": "NOT_ALIVE",
70
- "158": "NON_PARTICIPATING",
71
- "159": "NOT_ALIVE",
72
- "16": "NOT_ALIVE",
73
  "160": "NOT_ALIVE",
74
  "161": "NOT_ALIVE",
75
  "162": "NON_PARTICIPATING",
76
- "163": "NOT_ALIVE",
77
  "164": "NON_PARTICIPATING",
78
  "165": "NON_PARTICIPATING",
79
  "166": "NON_PARTICIPATING",
@@ -83,144 +83,144 @@
83
  "17": "NOT_ALIVE",
84
  "170": "NOT_ALIVE",
85
  "171": "NOT_ALIVE",
86
- "172": "NON_PARTICIPATING",
87
- "173": "NOT_ALIVE",
88
- "174": "NON_PARTICIPATING",
89
  "175": "NON_PARTICIPATING",
90
  "176": "NOT_ALIVE",
91
- "177": "NON_PARTICIPATING",
92
- "178": "NON_PARTICIPATING",
93
  "179": "NON_PARTICIPATING",
94
  "18": "NON_PARTICIPATING",
95
  "180": "NON_PARTICIPATING",
96
  "181": "NOT_ALIVE",
97
  "182": "NOT_ALIVE",
98
- "183": "NON_PARTICIPATING",
99
  "184": "NON_PARTICIPATING",
100
- "185": "NON_PARTICIPATING",
101
  "186": "NON_PARTICIPATING",
102
  "187": "NON_PARTICIPATING",
103
  "188": "NON_PARTICIPATING",
104
- "189": "NON_PARTICIPATING",
105
  "19": "NON_PARTICIPATING",
106
  "190": "NON_PARTICIPATING",
107
- "191": "NON_PARTICIPATING",
108
  "192": "NON_PARTICIPATING",
109
  "193": "NOT_ALIVE",
110
  "194": "NON_PARTICIPATING",
111
  "195": "NOT_ALIVE",
112
- "196": "NON_PARTICIPATING",
113
- "197": "NOT_ALIVE",
114
- "198": "NON_PARTICIPATING",
115
  "199": "NON_PARTICIPATING",
116
  "2": "NON_PARTICIPATING",
117
  "20": "NOT_ALIVE",
118
- "200": "NON_PARTICIPATING",
119
- "201": "NOT_ALIVE",
120
  "202": "NON_PARTICIPATING",
121
- "203": "NOT_ALIVE",
122
  "204": "NON_PARTICIPATING",
123
  "205": "NOT_ALIVE",
124
- "206": "NOT_ALIVE",
125
  "207": "NOT_ALIVE",
126
  "208": "NON_PARTICIPATING",
127
  "209": "NOT_ALIVE",
128
  "21": "NOT_ALIVE",
129
  "210": "NOT_ALIVE",
130
- "211": "NON_PARTICIPATING",
131
- "212": "NON_PARTICIPATING",
132
  "213": "NON_PARTICIPATING",
133
  "214": "NON_PARTICIPATING",
134
  "215": "NON_PARTICIPATING",
135
  "216": "NON_PARTICIPATING",
136
  "217": "NOT_ALIVE",
137
- "218": "NON_PARTICIPATING",
138
  "219": "NOT_ALIVE",
139
- "22": "NON_PARTICIPATING",
140
  "220": "NON_PARTICIPATING",
141
  "221": "NON_PARTICIPATING",
142
- "222": "NON_PARTICIPATING",
143
- "223": "NON_PARTICIPATING",
144
  "224": "NOT_ALIVE",
145
- "225": "NON_PARTICIPATING",
146
  "226": "NON_PARTICIPATING",
147
- "227": "NON_PARTICIPATING",
148
  "228": "NON_PARTICIPATING",
149
  "229": "NON_PARTICIPATING",
150
- "23": "NON_PARTICIPATING",
151
  "230": "NOT_ALIVE",
152
  "231": "NOT_ALIVE",
153
  "232": "NOT_ALIVE",
154
- "233": "NON_PARTICIPATING",
155
- "234": "NOT_ALIVE",
156
  "235": "NON_PARTICIPATING",
157
- "236": "NON_PARTICIPATING",
158
- "237": "NON_PARTICIPATING",
159
  "238": "NON_PARTICIPATING",
160
  "239": "NOT_ALIVE",
161
- "24": "NOT_ALIVE",
162
- "240": "NON_PARTICIPATING",
163
- "241": "NOT_ALIVE",
164
  "242": "NON_PARTICIPATING",
165
- "243": "NON_PARTICIPATING",
166
  "244": "NON_PARTICIPATING",
167
  "245": "NON_PARTICIPATING",
168
- "246": "NOT_ALIVE",
169
- "247": "NON_PARTICIPATING",
170
  "248": "NOT_ALIVE",
171
  "249": "NON_PARTICIPATING",
172
  "25": "SUCCESS",
173
- "250": "NOT_ALIVE",
174
  "251": "NOT_ALIVE",
175
  "252": "NOT_ALIVE",
176
- "253": "NON_PARTICIPATING",
177
  "254": "NOT_ALIVE",
178
- "255": "NOT_ALIVE",
179
- "26": "NON_PARTICIPATING",
180
  "27": "NOT_ALIVE",
181
- "28": "NON_PARTICIPATING",
182
  "29": "NOT_ALIVE",
183
  "3": "NON_PARTICIPATING",
184
- "30": "NOT_ALIVE",
185
  "31": "NON_PARTICIPATING",
186
  "32": "NON_PARTICIPATING",
187
  "33": "NOT_ALIVE",
188
- "34": "NOT_ALIVE",
189
  "35": "NON_PARTICIPATING",
190
- "36": "NOT_ALIVE",
191
- "37": "NON_PARTICIPATING",
192
  "38": "NOT_ALIVE",
193
  "39": "NON_PARTICIPATING",
194
- "4": "NON_PARTICIPATING",
195
- "40": "NON_PARTICIPATING",
196
  "41": "NON_PARTICIPATING",
197
- "42": "NOT_ALIVE",
198
  "43": "NOT_ALIVE",
199
- "44": "NON_PARTICIPATING",
200
  "45": "NON_PARTICIPATING",
201
- "46": "NON_PARTICIPATING",
202
- "47": "NON_PARTICIPATING",
203
- "48": "NOT_ALIVE",
204
  "49": "NOT_ALIVE",
205
- "5": "NON_PARTICIPATING",
206
  "50": "NON_PARTICIPATING",
207
- "51": "NON_PARTICIPATING",
208
- "52": "NON_PARTICIPATING",
209
  "53": "NON_PARTICIPATING",
210
- "54": "NOT_ALIVE",
211
- "55": "NOT_ALIVE",
212
  "56": "NON_PARTICIPATING",
213
- "57": "NOT_ALIVE",
214
  "58": "NON_PARTICIPATING",
215
- "59": "NOT_ALIVE",
216
- "6": "NON_PARTICIPATING",
217
- "60": "NOT_ALIVE",
218
  "61": "NON_PARTICIPATING",
219
- "62": "NON_PARTICIPATING",
220
  "63": "NON_PARTICIPATING",
221
- "64": "NOT_ALIVE",
222
- "65": "NON_PARTICIPATING",
223
- "66": "NON_PARTICIPATING",
224
  "67": "NOT_ALIVE",
225
  "68": "NON_PARTICIPATING",
226
  "69": "NOT_ALIVE",
@@ -229,34 +229,34 @@
229
  "71": "NON_PARTICIPATING",
230
  "72": "NON_PARTICIPATING",
231
  "73": "NON_PARTICIPATING",
232
- "74": "NOT_ALIVE",
233
  "75": "NOT_ALIVE",
234
  "76": "NOT_ALIVE",
235
- "77": "NON_PARTICIPATING",
236
- "78": "NON_PARTICIPATING",
237
- "79": "NON_PARTICIPATING",
238
  "8": "NOT_ALIVE",
239
  "80": "NON_PARTICIPATING",
240
  "81": "NOT_ALIVE",
241
- "82": "NON_PARTICIPATING",
242
  "83": "NOT_ALIVE",
243
- "84": "NON_PARTICIPATING",
244
  "85": "NOT_ALIVE",
245
  "86": "NON_PARTICIPATING",
246
  "87": "NOT_ALIVE",
247
  "88": "NOT_ALIVE",
248
  "89": "NOT_ALIVE",
249
  "9": "NON_PARTICIPATING",
250
- "90": "NOT_ALIVE",
251
- "91": "NON_PARTICIPATING",
252
- "92": "NOT_ALIVE",
253
- "93": "NOT_ALIVE",
254
- "94": "NOT_ALIVE",
255
- "95": "NOT_ALIVE",
256
- "96": "NOT_ALIVE",
257
  "97": "NON_PARTICIPATING",
258
  "98": "NOT_ALIVE",
259
- "99": "NON_PARTICIPATING"
260
  },
261
  "architectures": [
262
  "LlamaForCausalLM"
@@ -264,17 +264,17 @@
264
  "attention_bias": false,
265
  "attention_dropout": 0.0,
266
  "block_list": [
267
- 6087216,
268
- 6087242
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": 128,
276
  "intermediate_size": 5632,
277
- "last_allreduce_block": 6062756,
278
  "max_position_embeddings": 2048,
279
  "mlp_bias": false,
280
  "model_type": "llama",
 
6
  "10": "NON_PARTICIPATING",
7
  "100": "NOT_ALIVE",
8
  "101": "NOT_ALIVE",
9
+ "102": "NOT_ALIVE",
10
+ "103": "NOT_ALIVE",
11
  "104": "NOT_ALIVE",
12
  "105": "NON_PARTICIPATING",
13
  "106": "NON_PARTICIPATING",
14
+ "107": "NOT_ALIVE",
15
+ "108": "SUCCESS",
16
  "109": "NON_PARTICIPATING",
17
  "11": "NON_PARTICIPATING",
18
  "110": "NON_PARTICIPATING",
19
  "111": "NOT_ALIVE",
20
  "112": "NON_PARTICIPATING",
21
+ "113": "SUCCESS",
22
  "114": "NON_PARTICIPATING",
23
  "115": "NON_PARTICIPATING",
24
  "116": "NON_PARTICIPATING",
25
  "117": "NON_PARTICIPATING",
26
+ "118": "NON_PARTICIPATING",
27
  "119": "NOT_ALIVE",
28
  "12": "NOT_ALIVE",
29
+ "120": "NOT_ALIVE",
30
+ "121": "NOT_ALIVE",
31
  "122": "NON_PARTICIPATING",
32
  "123": "NON_PARTICIPATING",
33
+ "124": "NOT_ALIVE",
34
  "125": "NON_PARTICIPATING",
35
  "126": "NON_PARTICIPATING",
36
  "127": "NOT_ALIVE",
37
+ "128": "NOT_ALIVE",
38
  "129": "NON_PARTICIPATING",
39
+ "13": "NON_PARTICIPATING",
40
  "130": "NON_PARTICIPATING",
41
  "131": "NOT_ALIVE",
42
+ "132": "NOT_ALIVE",
43
  "133": "NON_PARTICIPATING",
44
  "134": "NON_PARTICIPATING",
45
  "135": "NON_PARTICIPATING",
46
  "136": "NON_PARTICIPATING",
47
  "137": "NON_PARTICIPATING",
48
+ "138": "SUCCESS",
49
+ "139": "NOT_ALIVE",
50
  "14": "NOT_ALIVE",
51
+ "140": "NON_PARTICIPATING",
52
  "141": "NON_PARTICIPATING",
53
  "142": "NOT_ALIVE",
54
+ "143": "NOT_ALIVE",
55
  "144": "NON_PARTICIPATING",
56
  "145": "NOT_ALIVE",
57
+ "146": "NON_PARTICIPATING",
58
  "147": "NON_PARTICIPATING",
59
  "148": "NON_PARTICIPATING",
60
+ "149": "NON_PARTICIPATING",
61
  "15": "NOT_ALIVE",
62
  "150": "NON_PARTICIPATING",
63
  "151": "NOT_ALIVE",
64
+ "152": "NOT_ALIVE",
65
  "153": "NOT_ALIVE",
66
  "154": "NON_PARTICIPATING",
67
+ "155": "NOT_ALIVE",
68
  "156": "NOT_ALIVE",
69
+ "157": "NON_PARTICIPATING",
70
+ "158": "NOT_ALIVE",
71
+ "159": "NON_PARTICIPATING",
72
+ "16": "NON_PARTICIPATING",
73
  "160": "NOT_ALIVE",
74
  "161": "NOT_ALIVE",
75
  "162": "NON_PARTICIPATING",
76
+ "163": "NON_PARTICIPATING",
77
  "164": "NON_PARTICIPATING",
78
  "165": "NON_PARTICIPATING",
79
  "166": "NON_PARTICIPATING",
 
83
  "17": "NOT_ALIVE",
84
  "170": "NOT_ALIVE",
85
  "171": "NOT_ALIVE",
86
+ "172": "NOT_ALIVE",
87
+ "173": "NON_PARTICIPATING",
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": "NON_PARTICIPATING",
95
  "180": "NON_PARTICIPATING",
96
  "181": "NOT_ALIVE",
97
  "182": "NOT_ALIVE",
98
+ "183": "SUCCESS",
99
  "184": "NON_PARTICIPATING",
100
+ "185": "NOT_ALIVE",
101
  "186": "NON_PARTICIPATING",
102
  "187": "NON_PARTICIPATING",
103
  "188": "NON_PARTICIPATING",
104
+ "189": "NOT_ALIVE",
105
  "19": "NON_PARTICIPATING",
106
  "190": "NON_PARTICIPATING",
107
+ "191": "NOT_ALIVE",
108
  "192": "NON_PARTICIPATING",
109
  "193": "NOT_ALIVE",
110
  "194": "NON_PARTICIPATING",
111
  "195": "NOT_ALIVE",
112
+ "196": "SUCCESS",
113
+ "197": "NON_PARTICIPATING",
114
+ "198": "NOT_ALIVE",
115
  "199": "NON_PARTICIPATING",
116
  "2": "NON_PARTICIPATING",
117
  "20": "NOT_ALIVE",
118
+ "200": "NOT_ALIVE",
119
+ "201": "NON_PARTICIPATING",
120
  "202": "NON_PARTICIPATING",
121
+ "203": "NON_PARTICIPATING",
122
  "204": "NON_PARTICIPATING",
123
  "205": "NOT_ALIVE",
124
+ "206": "NON_PARTICIPATING",
125
  "207": "NOT_ALIVE",
126
  "208": "NON_PARTICIPATING",
127
  "209": "NOT_ALIVE",
128
  "21": "NOT_ALIVE",
129
  "210": "NOT_ALIVE",
130
+ "211": "NOT_ALIVE",
131
+ "212": "NOT_ALIVE",
132
  "213": "NON_PARTICIPATING",
133
  "214": "NON_PARTICIPATING",
134
  "215": "NON_PARTICIPATING",
135
  "216": "NON_PARTICIPATING",
136
  "217": "NOT_ALIVE",
137
+ "218": "NOT_ALIVE",
138
  "219": "NOT_ALIVE",
139
+ "22": "NOT_ALIVE",
140
  "220": "NON_PARTICIPATING",
141
  "221": "NON_PARTICIPATING",
142
+ "222": "NOT_ALIVE",
143
+ "223": "NOT_ALIVE",
144
  "224": "NOT_ALIVE",
145
+ "225": "NOT_ALIVE",
146
  "226": "NON_PARTICIPATING",
147
+ "227": "NOT_ALIVE",
148
  "228": "NON_PARTICIPATING",
149
  "229": "NON_PARTICIPATING",
150
+ "23": "NOT_ALIVE",
151
  "230": "NOT_ALIVE",
152
  "231": "NOT_ALIVE",
153
  "232": "NOT_ALIVE",
154
+ "233": "NOT_ALIVE",
155
+ "234": "NON_PARTICIPATING",
156
  "235": "NON_PARTICIPATING",
157
+ "236": "NOT_ALIVE",
158
+ "237": "NOT_ALIVE",
159
  "238": "NON_PARTICIPATING",
160
  "239": "NOT_ALIVE",
161
+ "24": "NON_PARTICIPATING",
162
+ "240": "NOT_ALIVE",
163
+ "241": "NON_PARTICIPATING",
164
  "242": "NON_PARTICIPATING",
165
+ "243": "NOT_ALIVE",
166
  "244": "NON_PARTICIPATING",
167
  "245": "NON_PARTICIPATING",
168
+ "246": "NON_PARTICIPATING",
169
+ "247": "NOT_ALIVE",
170
  "248": "NOT_ALIVE",
171
  "249": "NON_PARTICIPATING",
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": "NOT_ALIVE",
181
+ "28": "NOT_ALIVE",
182
  "29": "NOT_ALIVE",
183
  "3": "NON_PARTICIPATING",
184
+ "30": "SUCCESS",
185
  "31": "NON_PARTICIPATING",
186
  "32": "NON_PARTICIPATING",
187
  "33": "NOT_ALIVE",
188
+ "34": "NON_PARTICIPATING",
189
  "35": "NON_PARTICIPATING",
190
+ "36": "NON_PARTICIPATING",
191
+ "37": "NOT_ALIVE",
192
  "38": "NOT_ALIVE",
193
  "39": "NON_PARTICIPATING",
194
+ "4": "NOT_ALIVE",
195
+ "40": "NOT_ALIVE",
196
  "41": "NON_PARTICIPATING",
197
+ "42": "NON_PARTICIPATING",
198
  "43": "NOT_ALIVE",
199
+ "44": "NOT_ALIVE",
200
  "45": "NON_PARTICIPATING",
201
+ "46": "NOT_ALIVE",
202
+ "47": "SUCCESS",
203
+ "48": "NON_PARTICIPATING",
204
  "49": "NOT_ALIVE",
205
+ "5": "NOT_ALIVE",
206
  "50": "NON_PARTICIPATING",
207
+ "51": "NOT_ALIVE",
208
+ "52": "NOT_ALIVE",
209
  "53": "NON_PARTICIPATING",
210
+ "54": "NON_PARTICIPATING",
211
+ "55": "NON_PARTICIPATING",
212
  "56": "NON_PARTICIPATING",
213
+ "57": "NON_PARTICIPATING",
214
  "58": "NON_PARTICIPATING",
215
+ "59": "NON_PARTICIPATING",
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": "NOT_ALIVE",
223
+ "66": "NOT_ALIVE",
224
  "67": "NOT_ALIVE",
225
  "68": "NON_PARTICIPATING",
226
  "69": "NOT_ALIVE",
 
229
  "71": "NON_PARTICIPATING",
230
  "72": "NON_PARTICIPATING",
231
  "73": "NON_PARTICIPATING",
232
+ "74": "NON_PARTICIPATING",
233
  "75": "NOT_ALIVE",
234
  "76": "NOT_ALIVE",
235
+ "77": "SUCCESS",
236
+ "78": "NOT_ALIVE",
237
+ "79": "NOT_ALIVE",
238
  "8": "NOT_ALIVE",
239
  "80": "NON_PARTICIPATING",
240
  "81": "NOT_ALIVE",
241
+ "82": "NOT_ALIVE",
242
  "83": "NOT_ALIVE",
243
+ "84": "NOT_ALIVE",
244
  "85": "NOT_ALIVE",
245
  "86": "NON_PARTICIPATING",
246
  "87": "NOT_ALIVE",
247
  "88": "NOT_ALIVE",
248
  "89": "NOT_ALIVE",
249
  "9": "NON_PARTICIPATING",
250
+ "90": "NON_PARTICIPATING",
251
+ "91": "NOT_ALIVE",
252
+ "92": "NON_PARTICIPATING",
253
+ "93": "NON_PARTICIPATING",
254
+ "94": "NON_PARTICIPATING",
255
+ "95": "NON_PARTICIPATING",
256
+ "96": "NON_PARTICIPATING",
257
  "97": "NON_PARTICIPATING",
258
  "98": "NOT_ALIVE",
259
+ "99": "NOT_ALIVE"
260
  },
261
  "architectures": [
262
  "LlamaForCausalLM"
 
264
  "attention_bias": false,
265
  "attention_dropout": 0.0,
266
  "block_list": [
267
+ 6087391,
268
+ 6087418
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": 6087213,
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 sha256:4dfb2f0e301a4be6d00edecbb1329bed620f5d766bfcdd03045e7a0a5ccf9d9e
3
  size 44585978
 
1
  version https://git-lfs.github.com/spec/v1
2
+ oid sha256:b7ac334d7e6bc1fe003ee2d124096361328c40b96f644053f6a1a942d3e8df6e
3
  size 44585978
inner_optimizer.pt CHANGED
@@ -1,3 +1,3 @@
1
  version https://git-lfs.github.com/spec/v1
2
- oid sha256:91c9c506586f5d806478c0ecb925a342dc4473ec56dc4ed8352cf17dedaf37b6
3
  size 8800564076
 
1
  version https://git-lfs.github.com/spec/v1
2
+ oid sha256:c1bbb66ce9ec94d58d5f510b6cbdea37cec28f5247c24f891785f98aa12bdaef
3
  size 8800564076
model.safetensors CHANGED
@@ -1,3 +1,3 @@
1
  version https://git-lfs.github.com/spec/v1
2
- oid sha256:dc490ebadacf9a0992c9a0e2aaaee1b2d780900b39d3f111ac83bb86760470ed
3
  size 4400216536
 
1
  version https://git-lfs.github.com/spec/v1
2
+ oid sha256:bf8048b604b227fe309caaaeb8194307a6e1278d70d0fd4b33db356d304b54f8
3
  size 4400216536