Hugging Face's logo

fwtan
/
llama-1.1b-mobilequant-w4a8-s1024-e60-sym-hf

PyTorch
Model card Files Community
imfinethx commited on
Commit
afaf3b5
·
1 Parent(s): 9fc7352

v1

Browse files
Files changed (11) hide show
  1. .gitattributes +1 -0
  2. act_dict.json +3 -0
  3. config.json +3 -0
  4. default_qcfg.json +3 -0
  5. generation_config.json +3 -0
  6. log_rank0_1722167032.txt +67 -0
  7. pytorch_model.bin +3 -0
  8. results.json +3 -0
  9. special_tokens_map.json +3 -0
  10. tokenizer.model +3 -0
  11. tokenizer_config.json +3 -0
.gitattributes CHANGED
@@ -33,3 +33,4 @@ saved_model/**/* filter=lfs diff=lfs merge=lfs -text
33
  *.zip filter=lfs diff=lfs merge=lfs -text
34
  *.zst filter=lfs diff=lfs merge=lfs -text
35
  *tfevents* filter=lfs diff=lfs merge=lfs -text
 
 
33
  *.zip filter=lfs diff=lfs merge=lfs -text
34
  *.zst filter=lfs diff=lfs merge=lfs -text
35
  *tfevents* filter=lfs diff=lfs merge=lfs -text
36
+ *.json filter=lfs diff=lfs merge=lfs -text
act_dict.json ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:d120f687650338fd6f86e47902bc9a5912525cf3c1ea42d271be8f4d7742d0ef
3
+ size 53107
config.json ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:3813577311a23b9ae7530b6b0504e7a6560aedbe813bcee72d479a6f0bead81e
3
+ size 1307
default_qcfg.json ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:ea04fde39fe700ec7a56865dc09a141590b042edd6ea2fb1fe27f5d9290cfeeb
3
+ size 140550
generation_config.json ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:56cf26bf2918fddebbfcb484144e503431a431f8b3695a502d4228b7978cd216
3
+ size 129
log_rank0_1722167032.txt ADDED
@@ -0,0 +1,67 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ [2024-07-28 11:43:52 root] (mobilequant.py 200): INFO Namespace(hf_path='checkpoints/hfmodels/llama-1.1b', dtype='float32', output_dir='results/llama-1.1b-e2e-w4a8-s1024-e60-sym', cache_dir='./cache', resume=None, calib_dataset='pile', nsamples=1024, seqlen=2048, act_dict_path='checkpoints/hfmodels/llama-1.1b/act_dict.json', override_qcfg_path='checkpoints/hfmodels/llama-1.1b/default_qcfg.json', weight_bitwidth=4, weight_group_size=-1, weight_is_per_channel=True, weight_is_symmetric=True, weight_is_dynamic=False, act_bitwidth=8, act_group_size=-1, act_is_per_channel=False, act_is_symmetric=False, act_is_dynamic=False, let=True, lwc=True, lrl=True, let_lr=0.001, lwc_lr=0.001, lrl_lr=1e-06, let_min_lr=0.0001, lwc_min_lr=0.0001, lrl_min_lr=1e-07, wd=0, epochs=60, warmup_epochs=0, use_shift=False, aug_loss=False, deactive_amp=True, eval_ppl=False, batch_size=1, num_fewshot=0, tasks='wikitext', mode='e2e', original_omniquant=False, cache_in_gpu=False, use_8bit_softmax_input=False, use_8bit_softmax_output=False, model_family='llama')
2
+ [2024-07-28 11:44:04 root] (mobilequant.py 295): INFO === start quantization ===
3
+ [2024-07-28 11:44:04 root] (mobilequant.py 301): INFO load calibration set from ./cache/dataloader_llama_pile_1024.cache
4
+ [2024-07-28 11:44:04 root] (algorithm.py 586): INFO Starting ...
5
+ [2024-07-28 12:15:24 root] (algorithm.py 757): INFO Epoch 0 loss:0.21866346895694733 norm:1681.2789306640625 max memory_allocated 36892.55322265625
6
+ [2024-07-28 12:44:10 root] (algorithm.py 757): INFO Epoch 1 loss:0.14636600017547607 norm:1114.6199951171875 max memory_allocated 36892.77587890625
7
+ [2024-07-28 13:12:47 root] (algorithm.py 757): INFO Epoch 2 loss:0.12111612409353256 norm:580.7744750976562 max memory_allocated 36892.77587890625
8
+ [2024-07-28 13:41:10 root] (algorithm.py 757): INFO Epoch 3 loss:0.10590877383947372 norm:248.77093505859375 max memory_allocated 36892.77587890625
9
+ [2024-07-28 14:09:36 root] (algorithm.py 757): INFO Epoch 4 loss:0.09518460929393768 norm:97.11088562011719 max memory_allocated 36892.77587890625
10
+ [2024-07-28 14:38:22 root] (algorithm.py 757): INFO Epoch 5 loss:0.08710672706365585 norm:40.954036712646484 max memory_allocated 36892.77587890625
11
+ [2024-07-28 15:06:56 root] (algorithm.py 757): INFO Epoch 6 loss:0.08285680413246155 norm:26.62285804748535 max memory_allocated 36892.77587890625
12
+ [2024-07-28 15:35:13 root] (algorithm.py 757): INFO Epoch 7 loss:0.07977326214313507 norm:21.728118896484375 max memory_allocated 36892.77587890625
13
+ [2024-07-28 16:03:26 root] (algorithm.py 757): INFO Epoch 8 loss:0.07773774862289429 norm:48.750118255615234 max memory_allocated 36892.77587890625
14
+ [2024-07-28 16:31:38 root] (algorithm.py 757): INFO Epoch 9 loss:0.07568226754665375 norm:16.35084342956543 max memory_allocated 36892.77587890625
15
+ [2024-07-28 16:59:52 root] (algorithm.py 757): INFO Epoch 10 loss:0.0739452913403511 norm:13.078861236572266 max memory_allocated 36892.77587890625
16
+ [2024-07-28 17:28:05 root] (algorithm.py 757): INFO Epoch 11 loss:0.07255983352661133 norm:9.205175399780273 max memory_allocated 36892.77587890625
17
+ [2024-07-28 17:56:20 root] (algorithm.py 757): INFO Epoch 12 loss:0.07101286947727203 norm:8.06998062133789 max memory_allocated 36892.77587890625
18
+ [2024-07-28 18:24:35 root] (algorithm.py 757): INFO Epoch 13 loss:0.06989651173353195 norm:7.019185543060303 max memory_allocated 36892.77587890625
19
+ [2024-07-28 18:52:47 root] (algorithm.py 757): INFO Epoch 14 loss:0.0688510313630104 norm:10.036494255065918 max memory_allocated 36892.77587890625
20
+ [2024-07-28 19:20:58 root] (algorithm.py 757): INFO Epoch 15 loss:0.06784027069807053 norm:6.388965606689453 max memory_allocated 36892.77587890625
21
+ [2024-07-28 19:49:08 root] (algorithm.py 757): INFO Epoch 16 loss:0.06705167144536972 norm:5.253536701202393 max memory_allocated 36892.77587890625
22
+ [2024-07-28 20:17:22 root] (algorithm.py 757): INFO Epoch 17 loss:0.06623184680938721 norm:3.946946382522583 max memory_allocated 36892.77587890625
23
+ [2024-07-28 20:45:59 root] (algorithm.py 757): INFO Epoch 18 loss:0.06553078442811966 norm:2.7513582706451416 max memory_allocated 36892.77587890625
24
+ [2024-07-28 21:14:34 root] (algorithm.py 757): INFO Epoch 19 loss:0.06490112841129303 norm:2.4710121154785156 max memory_allocated 36892.77587890625
25
+ [2024-07-28 21:43:10 root] (algorithm.py 757): INFO Epoch 20 loss:0.06423458456993103 norm:2.2626869678497314 max memory_allocated 36892.77587890625
26
+ [2024-07-28 22:11:44 root] (algorithm.py 757): INFO Epoch 21 loss:0.0640585869550705 norm:2.1731929779052734 max memory_allocated 36892.77587890625
27
+ [2024-07-28 22:40:02 root] (algorithm.py 757): INFO Epoch 22 loss:0.06378667801618576 norm:3.736574172973633 max memory_allocated 36892.77587890625
28
+ [2024-07-28 23:08:48 root] (algorithm.py 757): INFO Epoch 23 loss:0.06283513456583023 norm:3.736647605895996 max memory_allocated 36892.77587890625
29
+ [2024-07-28 23:37:31 root] (algorithm.py 757): INFO Epoch 24 loss:0.06282263994216919 norm:2.6267850399017334 max memory_allocated 36892.77587890625
30
+ [2024-07-29 00:06:02 root] (algorithm.py 757): INFO Epoch 25 loss:0.062358200550079346 norm:2.6600632667541504 max memory_allocated 36892.77587890625
31
+ [2024-07-29 00:34:51 root] (algorithm.py 757): INFO Epoch 26 loss:0.061927199363708496 norm:3.3202695846557617 max memory_allocated 36892.77587890625
32
+ [2024-07-29 01:03:27 root] (algorithm.py 757): INFO Epoch 27 loss:0.06155331805348396 norm:3.592175245285034 max memory_allocated 36892.77587890625
33
+ [2024-07-29 01:32:04 root] (algorithm.py 757): INFO Epoch 28 loss:0.06120344623923302 norm:5.199376583099365 max memory_allocated 36892.77587890625
34
+ [2024-07-29 02:00:55 root] (algorithm.py 757): INFO Epoch 29 loss:0.06065047159790993 norm:3.463374614715576 max memory_allocated 36892.77587890625
35
+ [2024-07-29 02:29:47 root] (algorithm.py 757): INFO Epoch 30 loss:0.06023940071463585 norm:3.317837715148926 max memory_allocated 36892.77587890625
36
+ [2024-07-29 02:58:16 root] (algorithm.py 757): INFO Epoch 31 loss:0.05992317944765091 norm:4.901540279388428 max memory_allocated 36892.77587890625
37
+ [2024-07-29 03:26:58 root] (algorithm.py 757): INFO Epoch 32 loss:0.05969621241092682 norm:5.476743698120117 max memory_allocated 36892.77587890625
38
+ [2024-07-29 03:55:38 root] (algorithm.py 757): INFO Epoch 33 loss:0.05939965322613716 norm:6.387513160705566 max memory_allocated 36892.77587890625
39
+ [2024-07-29 04:24:14 root] (algorithm.py 757): INFO Epoch 34 loss:0.05917934328317642 norm:8.606504440307617 max memory_allocated 36892.77587890625
40
+ [2024-07-29 04:52:57 root] (algorithm.py 757): INFO Epoch 35 loss:0.05899833142757416 norm:9.730156898498535 max memory_allocated 36892.77587890625
41
+ [2024-07-29 05:21:24 root] (algorithm.py 757): INFO Epoch 36 loss:0.058788809925317764 norm:10.266510009765625 max memory_allocated 36892.77587890625
42
+ [2024-07-29 05:49:47 root] (algorithm.py 757): INFO Epoch 37 loss:0.058650266379117966 norm:13.551891326904297 max memory_allocated 36892.77587890625
43
+ [2024-07-29 06:18:43 root] (algorithm.py 757): INFO Epoch 38 loss:0.058563947677612305 norm:14.321013450622559 max memory_allocated 36892.77587890625
44
+ [2024-07-29 06:47:33 root] (algorithm.py 757): INFO Epoch 39 loss:0.058407921344041824 norm:14.9265718460083 max memory_allocated 36892.77587890625
45
+ [2024-07-29 07:16:00 root] (algorithm.py 757): INFO Epoch 40 loss:0.058379001915454865 norm:15.940937995910645 max memory_allocated 36892.77587890625
46
+ [2024-07-29 07:44:11 root] (algorithm.py 757): INFO Epoch 41 loss:0.05822858214378357 norm:17.473941802978516 max memory_allocated 36892.77587890625
47
+ [2024-07-29 08:12:22 root] (algorithm.py 757): INFO Epoch 42 loss:0.05808005481958389 norm:18.38253402709961 max memory_allocated 36892.77587890625
48
+ [2024-07-29 08:40:34 root] (algorithm.py 757): INFO Epoch 43 loss:0.0579712875187397 norm:19.174419403076172 max memory_allocated 36892.77587890625
49
+ [2024-07-29 09:08:44 root] (algorithm.py 757): INFO Epoch 44 loss:0.05774536356329918 norm:18.174427032470703 max memory_allocated 36892.77587890625
50
+ [2024-07-29 09:36:56 root] (algorithm.py 757): INFO Epoch 45 loss:0.05756397917866707 norm:18.94085693359375 max memory_allocated 36892.77587890625
51
+ [2024-07-29 10:05:13 root] (algorithm.py 757): INFO Epoch 46 loss:0.05749411880970001 norm:19.616201400756836 max memory_allocated 36892.77587890625
52
+ [2024-07-29 10:33:26 root] (algorithm.py 757): INFO Epoch 47 loss:0.057377662509679794 norm:22.645709991455078 max memory_allocated 36892.77587890625
53
+ [2024-07-29 11:01:38 root] (algorithm.py 757): INFO Epoch 48 loss:0.057325199246406555 norm:21.537010192871094 max memory_allocated 36892.77587890625
54
+ [2024-07-29 11:29:48 root] (algorithm.py 757): INFO Epoch 49 loss:0.05735406652092934 norm:19.116300582885742 max memory_allocated 36892.77587890625
55
+ [2024-07-29 11:57:58 root] (algorithm.py 757): INFO Epoch 50 loss:0.057689398527145386 norm:3.942178249359131 max memory_allocated 36892.77587890625
56
+ [2024-07-29 12:26:09 root] (algorithm.py 757): INFO Epoch 51 loss:0.05762183666229248 norm:2.9464619159698486 max memory_allocated 36892.77587890625
57
+ [2024-07-29 12:54:19 root] (algorithm.py 757): INFO Epoch 52 loss:0.0575714036822319 norm:3.0514955520629883 max memory_allocated 36892.77587890625
58
+ [2024-07-29 13:22:31 root] (algorithm.py 757): INFO Epoch 53 loss:0.057542480528354645 norm:2.685452461242676 max memory_allocated 36892.77587890625
59
+ [2024-07-29 13:50:42 root] (algorithm.py 757): INFO Epoch 54 loss:0.05763385444879532 norm:2.8299198150634766 max memory_allocated 36892.77587890625
60
+ [2024-07-29 14:18:53 root] (algorithm.py 757): INFO Epoch 55 loss:0.057591512799263 norm:3.2381484508514404 max memory_allocated 36892.77587890625
61
+ [2024-07-29 14:47:02 root] (algorithm.py 757): INFO Epoch 56 loss:0.05752972513437271 norm:4.4083476066589355 max memory_allocated 36892.77587890625
62
+ [2024-07-29 15:15:16 root] (algorithm.py 757): INFO Epoch 57 loss:0.057428035885095596 norm:4.001595973968506 max memory_allocated 36892.77587890625
63
+ [2024-07-29 15:44:02 root] (algorithm.py 757): INFO Epoch 58 loss:0.05734267830848694 norm:4.042488098144531 max memory_allocated 36892.77587890625
64
+ [2024-07-29 16:12:47 root] (algorithm.py 757): INFO Epoch 59 loss:0.05734148249030113 norm:3.4370956420898438 max memory_allocated 36892.77587890625
65
+ [2024-07-29 16:12:52 root] (mobilequant.py 310): INFO 102527.6491343975
66
+ [2024-07-29 16:12:57 huggingface_hub.repocard] (repocard.py 107): WARNING Repo card metadata block was not found. Setting CardData to empty.
67
+ [2024-07-29 16:15:07 root] (mobilequant.py 149): INFO {'results': {'wikitext': {'word_perplexity': 17.516998697487463, 'byte_perplexity': 1.7081782163132995, 'bits_per_byte': 0.772458500941157}}, 'versions': {'wikitext': 1}, 'config': {'model': None, 'model_args': None, 'num_fewshot': 0, 'batch_size': 1, 'batch_sizes': [], 'device': None, 'no_cache': True, 'limit': None, 'bootstrap_iters': 100000, 'description_dict': None}}
pytorch_model.bin ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:a8ee21c7f83233dee4754a9e67c966290c708a1dbd2a526b6db472b59433cb4d
3
+ size 4401700922
results.json ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:d118e0b9323bffb3b60f03fcc891df2475c2d6e5c004bc6f7fd6096b4b0dd99f
3
+ size 577
special_tokens_map.json ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:82d96d7a9e6ced037f12394b7ea6a5b02e6ca87e0d11edaa8d60d9be857ce7db
3
+ size 551
tokenizer.model ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:9e556afd44213b6bd1be2b850ebbbd98f5481437a8021afaf58ee7fb1818d347
3
+ size 499723
tokenizer_config.json ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:6025dc3da07c0b33ed86a1f613b1b6a7d92d27822ee1d860d86812afa66c1b64
3
+ size 1410