[2024-08-18 19:06:55 root] (mobilequant.py 132): INFO Namespace(hf_path='checkpoints/hfmodels/gemma-2b', dtype='float32', output_dir='results/gemma-2b-e2e-w8a8-s1024-e60-2', cache_dir='./cache', resume=None, calib_dataset='pile', nsamples=1024, seqlen=2048, act_dict_path='checkpoints/hfmodels/gemma-2b/act_dict.json', override_qcfg_path='checkpoints/hfmodels/gemma-2b/default_qcfg.json', weight_bitwidth=8, weight_group_size=-1, weight_is_per_channel=False, weight_is_symmetric=False, 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=0.005, lrl_lr=1e-08, let_min_lr=0.001, lwc_min_lr=0.005, lrl_min_lr=1e-08, wd=0, epochs=60, warmup_epochs=0, use_shift=False, aug_loss=False, deactive_amp=True, 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='gemma')
[2024-08-18 19:07:22 root] (mobilequant.py 218): INFO === start quantization ===
[2024-08-18 19:07:22 root] (mobilequant.py 224): INFO load calibration set from ./cache/dataloader_gemma_pile_1024.cache
[2024-08-18 19:07:22 root] (algorithm.py 588): INFO Starting ...
[2024-08-18 19:32:13 root] (algorithm.py 759): INFO Epoch 0 loss:25.90846824645996 norm:160226.0625 max memory_allocated 31517.0673828125 
[2024-08-18 19:54:46 root] (algorithm.py 759): INFO Epoch 1 loss:19.347578048706055 norm:78570.5625 max memory_allocated 31517.4580078125 
[2024-08-18 20:17:18 root] (algorithm.py 759): INFO Epoch 2 loss:17.150110244750977 norm:49755.51171875 max memory_allocated 31517.4580078125 
[2024-08-18 20:39:51 root] (algorithm.py 759): INFO Epoch 3 loss:15.792439460754395 norm:54759.51171875 max memory_allocated 31517.4580078125 
[2024-08-18 21:02:24 root] (algorithm.py 759): INFO Epoch 4 loss:14.791868209838867 norm:57764.5 max memory_allocated 31517.4580078125 
[2024-08-18 21:24:56 root] (algorithm.py 759): INFO Epoch 5 loss:14.017207145690918 norm:51240.875 max memory_allocated 31517.4580078125 
[2024-08-18 21:47:40 root] (algorithm.py 759): INFO Epoch 6 loss:13.271882057189941 norm:52759.140625 max memory_allocated 31517.4580078125 
[2024-08-18 22:10:26 root] (algorithm.py 759): INFO Epoch 7 loss:12.6400146484375 norm:54853.9453125 max memory_allocated 31517.4580078125 
[2024-08-18 22:33:11 root] (algorithm.py 759): INFO Epoch 8 loss:12.088427543640137 norm:57650.1875 max memory_allocated 31517.4580078125 
[2024-08-18 22:55:56 root] (algorithm.py 759): INFO Epoch 9 loss:11.698744773864746 norm:53694.671875 max memory_allocated 31517.4580078125 
[2024-08-18 23:18:40 root] (algorithm.py 759): INFO Epoch 10 loss:11.29923152923584 norm:70695.4296875 max memory_allocated 31517.4580078125 
[2024-08-18 23:41:24 root] (algorithm.py 759): INFO Epoch 11 loss:11.042778015136719 norm:73705.796875 max memory_allocated 31517.4580078125 
[2024-08-19 00:04:09 root] (algorithm.py 759): INFO Epoch 12 loss:10.773776054382324 norm:58964.9296875 max memory_allocated 31517.4580078125 
[2024-08-19 00:26:53 root] (algorithm.py 759): INFO Epoch 13 loss:10.542635917663574 norm:58500.33203125 max memory_allocated 31517.4580078125 
[2024-08-19 00:49:38 root] (algorithm.py 759): INFO Epoch 14 loss:10.298208236694336 norm:60006.8046875 max memory_allocated 31517.4580078125 
[2024-08-19 01:12:16 root] (algorithm.py 759): INFO Epoch 15 loss:10.12488842010498 norm:66282.3125 max memory_allocated 31517.4580078125 
[2024-08-19 01:34:53 root] (algorithm.py 759): INFO Epoch 16 loss:10.032916069030762 norm:81933.4609375 max memory_allocated 31517.4580078125 
[2024-08-19 01:57:35 root] (algorithm.py 759): INFO Epoch 17 loss:9.848230361938477 norm:79033.1328125 max memory_allocated 31517.4580078125 
[2024-08-19 02:20:19 root] (algorithm.py 759): INFO Epoch 18 loss:9.750896453857422 norm:86391.25 max memory_allocated 31517.4580078125 
[2024-08-19 02:43:01 root] (algorithm.py 759): INFO Epoch 19 loss:9.706632614135742 norm:76601.6640625 max memory_allocated 31517.4580078125 
[2024-08-19 03:05:46 root] (algorithm.py 759): INFO Epoch 20 loss:9.60936450958252 norm:75337.484375 max memory_allocated 31517.4580078125 
[2024-08-19 03:28:31 root] (algorithm.py 759): INFO Epoch 21 loss:9.500767707824707 norm:69197.1875 max memory_allocated 31517.4580078125 
[2024-08-19 03:51:17 root] (algorithm.py 759): INFO Epoch 22 loss:9.39614486694336 norm:76400.203125 max memory_allocated 31517.4580078125 
[2024-08-19 04:13:49 root] (algorithm.py 759): INFO Epoch 23 loss:9.311173439025879 norm:63358.2578125 max memory_allocated 31517.4580078125 
[2024-08-19 04:36:33 root] (algorithm.py 759): INFO Epoch 24 loss:9.23372745513916 norm:77975.46875 max memory_allocated 31517.4580078125 
[2024-08-19 04:59:19 root] (algorithm.py 759): INFO Epoch 25 loss:9.138859748840332 norm:88843.640625 max memory_allocated 31517.4580078125 
[2024-08-19 05:22:03 root] (algorithm.py 759): INFO Epoch 26 loss:9.114325523376465 norm:95887.53125 max memory_allocated 31517.4580078125 
[2024-08-19 05:44:48 root] (algorithm.py 759): INFO Epoch 27 loss:9.11436653137207 norm:96310.46875 max memory_allocated 31517.4580078125 
[2024-08-19 06:07:21 root] (algorithm.py 759): INFO Epoch 28 loss:9.00890827178955 norm:106653.3515625 max memory_allocated 31517.4580078125 
[2024-08-19 06:29:54 root] (algorithm.py 759): INFO Epoch 29 loss:8.922462463378906 norm:108257.484375 max memory_allocated 31517.4580078125 
[2024-08-19 06:52:32 root] (algorithm.py 759): INFO Epoch 30 loss:8.893733024597168 norm:95072.96875 max memory_allocated 31517.4580078125 
[2024-08-19 07:15:05 root] (algorithm.py 759): INFO Epoch 31 loss:8.86312198638916 norm:85664.203125 max memory_allocated 31517.4580078125 
[2024-08-19 07:37:38 root] (algorithm.py 759): INFO Epoch 32 loss:8.800697326660156 norm:84694.7109375 max memory_allocated 31517.4580078125 
[2024-08-19 08:00:21 root] (algorithm.py 759): INFO Epoch 33 loss:8.736035346984863 norm:80500.3359375 max memory_allocated 31517.4580078125 
[2024-08-19 08:23:05 root] (algorithm.py 759): INFO Epoch 34 loss:8.719179153442383 norm:78551.7734375 max memory_allocated 31517.4580078125 
[2024-08-19 08:45:52 root] (algorithm.py 759): INFO Epoch 35 loss:8.697623252868652 norm:66414.5625 max memory_allocated 31517.4580078125 
[2024-08-19 09:08:36 root] (algorithm.py 759): INFO Epoch 36 loss:8.61997127532959 norm:74990.875 max memory_allocated 31517.4580078125 
[2024-08-19 09:31:14 root] (algorithm.py 759): INFO Epoch 37 loss:8.51863956451416 norm:78579.078125 max memory_allocated 31517.4580078125 
[2024-08-19 09:53:50 root] (algorithm.py 759): INFO Epoch 38 loss:8.553720474243164 norm:82302.9453125 max memory_allocated 31517.4580078125 
[2024-08-19 10:16:28 root] (algorithm.py 759): INFO Epoch 39 loss:8.456298828125 norm:76609.6171875 max memory_allocated 31517.4580078125 
[2024-08-19 10:39:13 root] (algorithm.py 759): INFO Epoch 40 loss:8.419332504272461 norm:59694.96875 max memory_allocated 31517.4580078125 
[2024-08-19 11:01:47 root] (algorithm.py 759): INFO Epoch 41 loss:8.369998931884766 norm:63071.09375 max memory_allocated 31517.4580078125 
[2024-08-19 11:24:25 root] (algorithm.py 759): INFO Epoch 42 loss:8.37739086151123 norm:85883.8046875 max memory_allocated 31517.4580078125 
[2024-08-19 11:47:06 root] (algorithm.py 759): INFO Epoch 43 loss:8.346087455749512 norm:66095.2109375 max memory_allocated 31517.4580078125 
[2024-08-19 12:09:51 root] (algorithm.py 759): INFO Epoch 44 loss:8.274130821228027 norm:60869.04296875 max memory_allocated 31517.4580078125 
[2024-08-19 12:32:32 root] (algorithm.py 759): INFO Epoch 45 loss:8.286489486694336 norm:75136.71875 max memory_allocated 31517.4580078125 
[2024-08-19 12:55:07 root] (algorithm.py 759): INFO Epoch 46 loss:8.291207313537598 norm:108988.09375 max memory_allocated 31517.4580078125 
[2024-08-19 13:17:42 root] (algorithm.py 759): INFO Epoch 47 loss:8.167930603027344 norm:91749.671875 max memory_allocated 31517.4580078125 
[2024-08-19 13:40:15 root] (algorithm.py 759): INFO Epoch 48 loss:8.167179107666016 norm:79381.25 max memory_allocated 31517.4580078125 
[2024-08-19 14:02:50 root] (algorithm.py 759): INFO Epoch 49 loss:8.184112548828125 norm:99687.1328125 max memory_allocated 31517.4580078125 
[2024-08-19 14:25:34 root] (algorithm.py 759): INFO Epoch 50 loss:8.112832069396973 norm:82443.359375 max memory_allocated 31517.4580078125 
[2024-08-19 14:48:16 root] (algorithm.py 759): INFO Epoch 51 loss:8.096149444580078 norm:90759.0625 max memory_allocated 31517.4580078125 
[2024-08-19 15:11:02 root] (algorithm.py 759): INFO Epoch 52 loss:7.97967529296875 norm:75396.375 max memory_allocated 31517.4580078125 
[2024-08-19 15:33:48 root] (algorithm.py 759): INFO Epoch 53 loss:8.050765991210938 norm:77746.140625 max memory_allocated 31517.4580078125 
[2024-08-19 15:56:33 root] (algorithm.py 759): INFO Epoch 54 loss:8.006107330322266 norm:85508.375 max memory_allocated 31517.4580078125 
[2024-08-19 16:19:19 root] (algorithm.py 759): INFO Epoch 55 loss:7.933506965637207 norm:84601.25 max memory_allocated 31517.4580078125 
[2024-08-19 16:42:04 root] (algorithm.py 759): INFO Epoch 56 loss:7.909237384796143 norm:89451.375 max memory_allocated 31517.4580078125 
[2024-08-19 17:04:50 root] (algorithm.py 759): INFO Epoch 57 loss:7.860087871551514 norm:76771.46875 max memory_allocated 31517.4580078125 
[2024-08-19 17:27:32 root] (algorithm.py 759): INFO Epoch 58 loss:7.896796703338623 norm:92898.765625 max memory_allocated 31517.4580078125 
[2024-08-19 17:50:14 root] (algorithm.py 759): INFO Epoch 59 loss:7.850639343261719 norm:71710.671875 max memory_allocated 31517.4580078125 
[2024-08-19 17:50:17 root] (mobilequant.py 233): INFO 81775.49729943275
[2024-08-19 17:50:23 huggingface_hub.repocard] (repocard.py 107): WARNING Repo card metadata block was not found. Setting CardData to empty.
[2024-08-19 17:54:12 root] (mobilequant.py 110): INFO {'results': {'wikitext': {'word_perplexity': 20.232104616802037, 'byte_perplexity': 1.7548346741959875, 'bits_per_byte': 0.8113351183061009}}, '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}}