cleanup: remove invalid ASVD baselines (wrong whitening)
Browse files
baselines/asvd_ppl_cached.log
DELETED
|
@@ -1,46 +0,0 @@
|
|
| 1 |
-
Loading model...
|
| 2 |
-
`torch_dtype` is deprecated! Use `dtype` instead!
|
| 3 |
-
2026-04-21 07:55:46.654229: I tensorflow/core/util/port.cc:153] oneDNN custom operations are on. You may see slightly different numerical results due to floating-point round-off errors from different computation orders. To turn them off, set the environment variable `TF_ENABLE_ONEDNN_OPTS=0`.
|
| 4 |
-
2026-04-21 07:55:46.674718: E external/local_xla/xla/stream_executor/cuda/cuda_fft.cc:467] Unable to register cuFFT factory: Attempting to register factory for plugin cuFFT when one has already been registered
|
| 5 |
-
WARNING: All log messages before absl::InitializeLog() is called are written to STDERR
|
| 6 |
-
E0000 00:00:1776758146.693305 2320089 cuda_dnn.cc:8579] Unable to register cuDNN factory: Attempting to register factory for plugin cuDNN when one has already been registered
|
| 7 |
-
E0000 00:00:1776758146.699392 2320089 cuda_blas.cc:1407] Unable to register cuBLAS factory: Attempting to register factory for plugin cuBLAS when one has already been registered
|
| 8 |
-
W0000 00:00:1776758146.714805 2320089 computation_placer.cc:177] computation placer already registered. Please check linkage and avoid linking the same target more than once.
|
| 9 |
-
W0000 00:00:1776758146.714830 2320089 computation_placer.cc:177] computation placer already registered. Please check linkage and avoid linking the same target more than once.
|
| 10 |
-
W0000 00:00:1776758146.714834 2320089 computation_placer.cc:177] computation placer already registered. Please check linkage and avoid linking the same target more than once.
|
| 11 |
-
W0000 00:00:1776758146.714837 2320089 computation_placer.cc:177] computation placer already registered. Please check linkage and avoid linking the same target more than once.
|
| 12 |
-
2026-04-21 07:55:46.718838: I tensorflow/core/platform/cpu_feature_guard.cc:210] This TensorFlow binary is optimized to use available CPU instructions in performance-critical operations.
|
| 13 |
-
To enable the following instructions: AVX2 AVX512F AVX512_VNNI FMA, in other operations, rebuild TensorFlow with the appropriate compiler flags.
|
| 14 |
-
AttributeError: 'MessageFactory' object has no attribute 'GetPrototype'
|
| 15 |
-
AttributeError: 'MessageFactory' object has no attribute 'GetPrototype'
|
| 16 |
-
AttributeError: 'MessageFactory' object has no attribute 'GetPrototype'
|
| 17 |
-
AttributeError: 'MessageFactory' object has no attribute 'GetPrototype'
|
| 18 |
-
AttributeError: 'MessageFactory' object has no attribute 'GetPrototype'
|
| 19 |
-
|
| 20 |
-
Collecting covariances...
|
| 21 |
-
Calibration: 16/256 samples
|
| 22 |
-
Calibration: 32/256 samples
|
| 23 |
-
Calibration: 48/256 samples
|
| 24 |
-
Calibration: 64/256 samples
|
| 25 |
-
Calibration: 80/256 samples
|
| 26 |
-
Calibration: 96/256 samples
|
| 27 |
-
Calibration: 112/256 samples
|
| 28 |
-
Calibration: 128/256 samples
|
| 29 |
-
Calibration: 144/256 samples
|
| 30 |
-
Calibration: 160/256 samples
|
| 31 |
-
Calibration: 176/256 samples
|
| 32 |
-
Calibration: 192/256 samples
|
| 33 |
-
Calibration: 208/256 samples
|
| 34 |
-
Calibration: 224/256 samples
|
| 35 |
-
Calibration: 240/256 samples
|
| 36 |
-
Calibration: 256/256 samples
|
| 37 |
-
|
| 38 |
-
============================================================
|
| 39 |
-
ASVD PPL-mode at ratio=0.4
|
| 40 |
-
[ASVD-PPL] Caching block inputs (32 blocks × 32 samples)...
|
| 41 |
-
[ASVD-PPL] Cache ready. Shapes: torch.Size([32, 2048, 4096]) per block
|
| 42 |
-
[ASVD-PPL] Validation: full=5.3369 cached=5.3368 diff=0.000104
|
| 43 |
-
[ASVD-PPL] Measuring sensitivity: 224 layers × 6 ratios
|
| 44 |
-
[1/224] self_attn.q_proj layer=0 pr=0.4 ppl=5.34
|
| 45 |
-
[1/224] self_attn.q_proj layer=0 pr=0.5 ppl=5.34
|
| 46 |
-
[1/224] self_attn.q_proj layer=0 pr=0.6 ppl=5.34
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|