MBA4215 commited on
Commit
fa5a1d3
·
verified ·
1 Parent(s): ff06498

Run 5. Outer Step 50. Inner Step 2.

Browse files
Files changed (4) hide show
  1. config.json +94 -103
  2. generation_config.json +7 -0
  3. inner_optimizer.pt +2 -2
  4. model.safetensors +2 -2
config.json CHANGED
@@ -1,16 +1,15 @@
1
  {
2
- "_name_or_path": "distributed/optimized-gpt2-1b",
3
- "activation_function": "gelu_new",
4
  "all_reduce_scores": {
5
  "0": "NON_PARTICIPATING",
6
- "1": "NON_PARTICIPATING",
7
- "10": "NON_PARTICIPATING",
8
  "100": "NON_PARTICIPATING",
9
  "101": "NON_PARTICIPATING",
10
  "102": "NON_PARTICIPATING",
11
  "103": "NON_PARTICIPATING",
12
  "104": "NON_PARTICIPATING",
13
- "105": "SUCCESS",
14
  "106": "NON_PARTICIPATING",
15
  "107": "NON_PARTICIPATING",
16
  "108": "NON_PARTICIPATING",
@@ -29,7 +28,7 @@
29
  "12": "SUCCESS",
30
  "120": "NON_PARTICIPATING",
31
  "121": "NON_PARTICIPATING",
32
- "122": "SUCCESS",
33
  "123": "NON_PARTICIPATING",
34
  "124": "NON_PARTICIPATING",
35
  "125": "NON_PARTICIPATING",
@@ -59,14 +58,14 @@
59
  "147": "SUCCESS",
60
  "148": "NON_PARTICIPATING",
61
  "149": "NON_PARTICIPATING",
62
- "15": "NON_PARTICIPATING",
63
  "150": "NON_PARTICIPATING",
64
  "151": "NON_PARTICIPATING",
65
  "152": "NON_PARTICIPATING",
66
  "153": "NON_PARTICIPATING",
67
  "154": "NON_PARTICIPATING",
68
  "155": "NON_PARTICIPATING",
69
- "156": "SUCCESS",
70
  "157": "NON_PARTICIPATING",
71
  "158": "NON_PARTICIPATING",
72
  "159": "NON_PARTICIPATING",
@@ -81,15 +80,15 @@
81
  "167": "NON_PARTICIPATING",
82
  "168": "NON_PARTICIPATING",
83
  "169": "NON_PARTICIPATING",
84
- "17": "NON_PARTICIPATING",
85
  "170": "NON_PARTICIPATING",
86
  "171": "NON_PARTICIPATING",
87
  "172": "NON_PARTICIPATING",
88
  "173": "NON_PARTICIPATING",
89
  "174": "NON_PARTICIPATING",
90
- "175": "SUCCESS",
91
  "176": "NON_PARTICIPATING",
92
- "177": "SUCCESS",
93
  "178": "NON_PARTICIPATING",
94
  "179": "NON_PARTICIPATING",
95
  "18": "SUCCESS",
@@ -100,39 +99,39 @@
100
  "184": "NON_PARTICIPATING",
101
  "185": "NON_PARTICIPATING",
102
  "186": "NON_PARTICIPATING",
103
- "187": "NON_PARTICIPATING",
104
  "188": "NON_PARTICIPATING",
105
  "189": "NON_PARTICIPATING",
106
- "19": "NON_PARTICIPATING",
107
  "190": "NON_PARTICIPATING",
108
  "191": "NON_PARTICIPATING",
109
  "192": "NON_PARTICIPATING",
110
  "193": "NON_PARTICIPATING",
111
- "194": "SUCCESS",
112
  "195": "NON_PARTICIPATING",
113
  "196": "NON_PARTICIPATING",
114
  "197": "NON_PARTICIPATING",
115
  "198": "NON_PARTICIPATING",
116
- "199": "NON_PARTICIPATING",
117
- "2": "NON_PARTICIPATING",
118
  "20": "SUCCESS",
119
- "200": "NON_PARTICIPATING",
120
  "201": "NON_PARTICIPATING",
121
  "202": "NON_PARTICIPATING",
122
- "203": "SUCCESS",
123
  "204": "NON_PARTICIPATING",
124
- "205": "SUCCESS",
125
  "206": "NON_PARTICIPATING",
126
  "207": "NON_PARTICIPATING",
127
  "208": "NON_PARTICIPATING",
128
  "209": "NON_PARTICIPATING",
129
- "21": "NON_PARTICIPATING",
130
- "210": "NON_PARTICIPATING",
131
  "211": "NON_PARTICIPATING",
132
- "212": "SUCCESS",
133
  "213": "NON_PARTICIPATING",
134
  "214": "NON_PARTICIPATING",
135
- "215": "SUCCESS",
136
  "216": "NON_PARTICIPATING",
137
  "217": "NON_PARTICIPATING",
138
  "218": "NON_PARTICIPATING",
@@ -140,15 +139,15 @@
140
  "22": "SUCCESS",
141
  "220": "NON_PARTICIPATING",
142
  "221": "NON_PARTICIPATING",
143
- "222": "SUCCESS",
144
  "223": "NON_PARTICIPATING",
145
  "224": "NON_PARTICIPATING",
146
- "225": "SUCCESS",
147
  "226": "NON_PARTICIPATING",
148
  "227": "NON_PARTICIPATING",
149
  "228": "NON_PARTICIPATING",
150
  "229": "NON_PARTICIPATING",
151
- "23": "NON_PARTICIPATING",
152
  "230": "NON_PARTICIPATING",
153
  "231": "NON_PARTICIPATING",
154
  "232": "NON_PARTICIPATING",
@@ -158,13 +157,13 @@
158
  "236": "NON_PARTICIPATING",
159
  "237": "NON_PARTICIPATING",
160
  "238": "NON_PARTICIPATING",
161
- "239": "SUCCESS",
162
  "24": "SUCCESS",
163
  "240": "NON_PARTICIPATING",
164
  "241": "NON_PARTICIPATING",
165
  "242": "NON_PARTICIPATING",
166
- "243": "SUCCESS",
167
- "244": "SUCCESS",
168
  "245": "NON_PARTICIPATING",
169
  "246": "NON_PARTICIPATING",
170
  "247": "NON_PARTICIPATING",
@@ -176,127 +175,119 @@
176
  "252": "NON_PARTICIPATING",
177
  "253": "NON_PARTICIPATING",
178
  "254": "NON_PARTICIPATING",
179
- "255": "SUCCESS",
180
- "26": "SUCCESS",
181
  "27": "SUCCESS",
182
  "28": "SUCCESS",
183
  "29": "SUCCESS",
184
- "3": "NON_PARTICIPATING",
185
- "30": "NON_PARTICIPATING",
186
- "31": "NON_PARTICIPATING",
187
  "32": "SUCCESS",
188
  "33": "NON_PARTICIPATING",
189
- "34": "NON_PARTICIPATING",
190
  "35": "SUCCESS",
191
- "36": "NON_PARTICIPATING",
192
  "37": "SUCCESS",
193
- "38": "NON_PARTICIPATING",
194
- "39": "NON_PARTICIPATING",
195
  "4": "SUCCESS",
196
  "40": "SUCCESS",
197
  "41": "SUCCESS",
198
- "42": "NON_PARTICIPATING",
199
  "43": "SUCCESS",
200
  "44": "SUCCESS",
201
  "45": "SUCCESS",
202
- "46": "NON_PARTICIPATING",
203
  "47": "SUCCESS",
204
- "48": "NON_PARTICIPATING",
205
  "49": "SUCCESS",
206
  "5": "SUCCESS",
207
  "50": "SUCCESS",
208
  "51": "SUCCESS",
209
- "52": "NON_PARTICIPATING",
210
- "53": "NON_PARTICIPATING",
211
  "54": "SUCCESS",
212
- "55": "NON_PARTICIPATING",
213
- "56": "NON_PARTICIPATING",
214
- "57": "NON_PARTICIPATING",
215
- "58": "NON_PARTICIPATING",
216
- "59": "NON_PARTICIPATING",
217
- "6": "NON_PARTICIPATING",
218
  "60": "SUCCESS",
219
  "61": "SUCCESS",
220
  "62": "SUCCESS",
221
  "63": "SUCCESS",
222
  "64": "SUCCESS",
223
- "65": "NON_PARTICIPATING",
224
- "66": "NON_PARTICIPATING",
225
- "67": "NON_PARTICIPATING",
226
- "68": "NON_PARTICIPATING",
227
  "69": "SUCCESS",
228
- "7": "NON_PARTICIPATING",
229
- "70": "NON_PARTICIPATING",
230
  "71": "SUCCESS",
231
  "72": "NON_PARTICIPATING",
232
- "73": "NON_PARTICIPATING",
233
  "74": "SUCCESS",
234
- "75": "NON_PARTICIPATING",
235
- "76": "NON_PARTICIPATING",
236
- "77": "SUCCESS",
237
- "78": "NON_PARTICIPATING",
238
- "79": "NON_PARTICIPATING",
239
  "8": "SUCCESS",
240
- "80": "NON_PARTICIPATING",
241
  "81": "SUCCESS",
242
- "82": "NON_PARTICIPATING",
243
  "83": "SUCCESS",
244
  "84": "SUCCESS",
245
  "85": "SUCCESS",
246
- "86": "NON_PARTICIPATING",
247
  "87": "SUCCESS",
248
  "88": "SUCCESS",
249
  "89": "SUCCESS",
250
  "9": "SUCCESS",
251
- "90": "NON_PARTICIPATING",
252
  "91": "SUCCESS",
253
- "92": "NON_PARTICIPATING",
254
- "93": "NON_PARTICIPATING",
255
- "94": "NON_PARTICIPATING",
256
  "95": "SUCCESS",
257
  "96": "SUCCESS",
258
- "97": "NON_PARTICIPATING",
259
  "98": "SUCCESS",
260
- "99": "NON_PARTICIPATING"
261
  },
262
  "architectures": [
263
- "GPTOptim"
264
  ],
265
- "attn_pdrop": 0.1,
266
- "auto_map": {
267
- "AutoConfig": "distributed/optimized-gpt2-500m--configuration_gpt_optimized.GPTOptimConfig",
268
- "AutoModelForCausalLM": "distributed/optimized-gpt2-500m--modeling_gpt_optimized.GPTOptim"
269
- },
270
  "block_list": [
271
- 5644604,
272
- 5644636
273
  ],
274
- "block_size": 1024,
275
- "bos_token_id": 50256,
276
- "embd_pdrop": 0.1,
277
- "eos_token_id": 50256,
278
  "initializer_range": 0.02,
279
- "inner_step": 10,
280
- "inner_steps": 0,
281
- "last_allreduce_block": 5644286,
282
- "layer_norm_epsilon": 1e-05,
283
- "model_type": "gpt_optimized",
284
- "n_embd": 1280,
285
- "n_head": 32,
286
- "n_inner": null,
287
- "n_layer": 48,
288
- "n_positions": 1024,
289
- "reorder_and_upcast_attn": false,
290
- "resid_pdrop": 0.1,
291
- "scale_attn_by_inverse_layer_idx": false,
292
- "scale_attn_weights": true,
293
- "summary_activation": null,
294
- "summary_first_dropout": 0.1,
295
- "summary_proj_to_labels": true,
296
- "summary_type": "cls_index",
297
- "summary_use_proj": true,
298
  "torch_dtype": "float32",
299
  "transformers_version": "4.39.3",
300
- "use_cache": true,
301
- "vocab_size": 50257
302
  }
 
1
  {
2
+ "_name_or_path": "distributed/llama-1b",
 
3
  "all_reduce_scores": {
4
  "0": "NON_PARTICIPATING",
5
+ "1": "SUCCESS",
6
+ "10": "SUCCESS",
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",
 
28
  "12": "SUCCESS",
29
  "120": "NON_PARTICIPATING",
30
  "121": "NON_PARTICIPATING",
31
+ "122": "NON_PARTICIPATING",
32
  "123": "NON_PARTICIPATING",
33
  "124": "NON_PARTICIPATING",
34
  "125": "NON_PARTICIPATING",
 
58
  "147": "SUCCESS",
59
  "148": "NON_PARTICIPATING",
60
  "149": "NON_PARTICIPATING",
61
+ "15": "SUCCESS",
62
  "150": "NON_PARTICIPATING",
63
  "151": "NON_PARTICIPATING",
64
  "152": "NON_PARTICIPATING",
65
  "153": "NON_PARTICIPATING",
66
  "154": "NON_PARTICIPATING",
67
  "155": "NON_PARTICIPATING",
68
+ "156": "NON_PARTICIPATING",
69
  "157": "NON_PARTICIPATING",
70
  "158": "NON_PARTICIPATING",
71
  "159": "NON_PARTICIPATING",
 
80
  "167": "NON_PARTICIPATING",
81
  "168": "NON_PARTICIPATING",
82
  "169": "NON_PARTICIPATING",
83
+ "17": "SUCCESS",
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",
 
99
  "184": "NON_PARTICIPATING",
100
  "185": "NON_PARTICIPATING",
101
  "186": "NON_PARTICIPATING",
102
+ "187": "SUCCESS",
103
  "188": "NON_PARTICIPATING",
104
  "189": "NON_PARTICIPATING",
105
+ "19": "SUCCESS",
106
  "190": "NON_PARTICIPATING",
107
  "191": "NON_PARTICIPATING",
108
  "192": "NON_PARTICIPATING",
109
  "193": "NON_PARTICIPATING",
110
+ "194": "NON_PARTICIPATING",
111
  "195": "NON_PARTICIPATING",
112
  "196": "NON_PARTICIPATING",
113
  "197": "NON_PARTICIPATING",
114
  "198": "NON_PARTICIPATING",
115
+ "199": "SUCCESS",
116
+ "2": "SUCCESS",
117
  "20": "SUCCESS",
118
+ "200": "SUCCESS",
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": "SUCCESS",
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": "NON_PARTICIPATING",
137
  "218": "NON_PARTICIPATING",
 
139
  "22": "SUCCESS",
140
  "220": "NON_PARTICIPATING",
141
  "221": "NON_PARTICIPATING",
142
+ "222": "NON_PARTICIPATING",
143
  "223": "NON_PARTICIPATING",
144
  "224": "NON_PARTICIPATING",
145
+ "225": "NON_PARTICIPATING",
146
  "226": "NON_PARTICIPATING",
147
  "227": "NON_PARTICIPATING",
148
  "228": "NON_PARTICIPATING",
149
  "229": "NON_PARTICIPATING",
150
+ "23": "SUCCESS",
151
  "230": "NON_PARTICIPATING",
152
  "231": "NON_PARTICIPATING",
153
  "232": "NON_PARTICIPATING",
 
157
  "236": "NON_PARTICIPATING",
158
  "237": "NON_PARTICIPATING",
159
  "238": "NON_PARTICIPATING",
160
+ "239": "NON_PARTICIPATING",
161
  "24": "SUCCESS",
162
  "240": "NON_PARTICIPATING",
163
  "241": "NON_PARTICIPATING",
164
  "242": "NON_PARTICIPATING",
165
+ "243": "NON_PARTICIPATING",
166
+ "244": "NON_PARTICIPATING",
167
  "245": "NON_PARTICIPATING",
168
  "246": "NON_PARTICIPATING",
169
  "247": "NON_PARTICIPATING",
 
175
  "252": "NON_PARTICIPATING",
176
  "253": "NON_PARTICIPATING",
177
  "254": "NON_PARTICIPATING",
178
+ "255": "NON_PARTICIPATING",
179
+ "26": "NON_PARTICIPATING",
180
  "27": "SUCCESS",
181
  "28": "SUCCESS",
182
  "29": "SUCCESS",
183
+ "3": "SUCCESS",
184
+ "30": "SUCCESS",
185
+ "31": "SUCCESS",
186
  "32": "SUCCESS",
187
  "33": "NON_PARTICIPATING",
188
+ "34": "SUCCESS",
189
  "35": "SUCCESS",
190
+ "36": "SUCCESS",
191
  "37": "SUCCESS",
192
+ "38": "SUCCESS",
193
+ "39": "SUCCESS",
194
  "4": "SUCCESS",
195
  "40": "SUCCESS",
196
  "41": "SUCCESS",
197
+ "42": "SUCCESS",
198
  "43": "SUCCESS",
199
  "44": "SUCCESS",
200
  "45": "SUCCESS",
201
+ "46": "SUCCESS",
202
  "47": "SUCCESS",
203
+ "48": "SUCCESS",
204
  "49": "SUCCESS",
205
  "5": "SUCCESS",
206
  "50": "SUCCESS",
207
  "51": "SUCCESS",
208
+ "52": "SUCCESS",
209
+ "53": "SUCCESS",
210
  "54": "SUCCESS",
211
+ "55": "SUCCESS",
212
+ "56": "SUCCESS",
213
+ "57": "SUCCESS",
214
+ "58": "SUCCESS",
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": "SUCCESS",
224
+ "67": "SUCCESS",
225
+ "68": "SUCCESS",
226
  "69": "SUCCESS",
227
+ "7": "SUCCESS",
228
+ "70": "SUCCESS",
229
  "71": "SUCCESS",
230
  "72": "NON_PARTICIPATING",
231
+ "73": "SUCCESS",
232
  "74": "SUCCESS",
233
+ "75": "SUCCESS",
234
+ "76": "SUCCESS",
235
+ "77": "NON_PARTICIPATING",
236
+ "78": "SUCCESS",
237
+ "79": "SUCCESS",
238
  "8": "SUCCESS",
239
+ "80": "SUCCESS",
240
  "81": "SUCCESS",
241
+ "82": "SUCCESS",
242
  "83": "SUCCESS",
243
  "84": "SUCCESS",
244
  "85": "SUCCESS",
245
+ "86": "SUCCESS",
246
  "87": "SUCCESS",
247
  "88": "SUCCESS",
248
  "89": "SUCCESS",
249
  "9": "SUCCESS",
250
+ "90": "SUCCESS",
251
  "91": "SUCCESS",
252
+ "92": "SUCCESS",
253
+ "93": "SUCCESS",
254
+ "94": "SUCCESS",
255
  "95": "SUCCESS",
256
  "96": "SUCCESS",
257
+ "97": "SUCCESS",
258
  "98": "SUCCESS",
259
+ "99": "SUCCESS"
260
  },
261
  "architectures": [
262
+ "LlamaForCausalLM"
263
  ],
264
+ "attention_bias": false,
265
+ "attention_dropout": 0.0,
 
 
 
266
  "block_list": [
267
+ 5773711,
268
+ 5773741
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": 5770468,
278
+ "max_position_embeddings": 2048,
279
+ "mlp_bias": false,
280
+ "model_type": "llama",
281
+ "num_attention_heads": 32,
282
+ "num_hidden_layers": 22,
283
+ "num_key_value_heads": 4,
284
+ "pretraining_tp": 1,
285
+ "rms_norm_eps": 1e-05,
286
+ "rope_scaling": null,
287
+ "rope_theta": 10000.0,
288
+ "tie_word_embeddings": false,
 
 
 
 
 
289
  "torch_dtype": "float32",
290
  "transformers_version": "4.39.3",
291
+ "use_cache": false,
292
+ "vocab_size": 32000
293
  }
generation_config.json ADDED
@@ -0,0 +1,7 @@
 
 
 
 
 
 
 
 
1
+ {
2
+ "_from_model_config": true,
3
+ "bos_token_id": 1,
4
+ "eos_token_id": 2,
5
+ "transformers_version": "4.39.3",
6
+ "use_cache": false
7
+ }
inner_optimizer.pt CHANGED
@@ -1,3 +1,3 @@
1
  version https://git-lfs.github.com/spec/v1
2
- oid sha256:6f525b8fbfc3d607911b1c7e6bb9b1ac20eefa3f9cd64771f600cf69275bcbc1
3
- size 8081782026
 
1
  version https://git-lfs.github.com/spec/v1
2
+ oid sha256:2043e6ac904ce9b06fd647fbb56e59617c599f65584a3abba347b6204cc0ada0
3
+ size 8800564076
model.safetensors CHANGED
@@ -1,3 +1,3 @@
1
  version https://git-lfs.github.com/spec/v1
2
- oid sha256:833cb990b156e19aaebdd407375d163425038ee942c400447850617ca3095f8f
3
- size 4040701744
 
1
  version https://git-lfs.github.com/spec/v1
2
+ oid sha256:e0b466a6b5cf80b25e1b041241da92e58af041e662f7ce7ae18cb47ee9f76029
3
+ size 4400216536