diff options
author | Richard Burton <richard.burton@arm.com> | 2021-08-12 17:26:30 +0100 |
---|---|---|
committer | Richard Burton <richard.burton@arm.com> | 2021-08-12 17:26:30 +0100 |
commit | 0d110594b8a50ce3311be5187f01de2e3b8fe995 (patch) | |
tree | 1e56414f491f1bbd29df4912e2354ac5e1682133 /docs | |
parent | d2b9853ca848f11dee55beedbb9d650763b3ed53 (diff) | |
download | ml-embedded-evaluation-kit-0d110594b8a50ce3311be5187f01de2e3b8fe995.tar.gz |
MLECO-1904: Update to use latest TFLu
* Now uses seperate TFLu github repo
* Fixes to align with API changes
* Update ASR model ops and re-enable ASR inference tests
* Set default release level to release_with_logs
Signed-off-by: Richard Burton <richard.burton@arm.com>
Change-Id: I57612088985dece1413c5c00a6e442381e07dd91
Diffstat (limited to 'docs')
-rw-r--r-- | docs/quick_start.md | 12 | ||||
-rw-r--r-- | docs/use_cases/img_class.md | 48 |
2 files changed, 30 insertions, 30 deletions
diff --git a/docs/quick_start.md b/docs/quick_start.md index d1039fe..878bdcf 100644 --- a/docs/quick_start.md +++ b/docs/quick_start.md @@ -78,11 +78,11 @@ curl -L https://github.com/ARM-software/ML-zoo/raw/1a92aa08c0de49a7304e0a7f3f59d --output ./resources_downloaded/asr/ifm0.npy curl -L https://github.com/ARM-software/ML-zoo/raw/1a92aa08c0de49a7304e0a7f3f59df6f4fd33ac8/models/speech_recognition/wav2letter/tflite_pruned_int8/testing_output/Identity_int8/0.npy \ --output ./resources_downloaded/asr/ofm0.npy -curl -L https://github.com/ARM-software/ML-zoo/raw/68b5fbc77ed28e67b2efc915997ea4477c1d9d5b/models/image_classification/mobilenet_v2_1.0_224/tflite_uint8/mobilenet_v2_1.0_224_quantized_1_default_1.tflite \ - --output ./resources_downloaded/img_class/mobilenet_v2_1.0_224_quantized_1_default_1.tflite -curl -L https://github.com/ARM-software/ML-zoo/raw/68b5fbc77ed28e67b2efc915997ea4477c1d9d5b/models/image_classification/mobilenet_v2_1.0_224/tflite_uint8/testing_input/input/0.npy \ +curl -L https://github.com/ARM-software/ML-zoo/raw/e0aa361b03c738047b9147d1a50e3f2dcb13dbcb/models/image_classification/mobilenet_v2_1.0_224/tflite_int8/mobilenet_v2_1.0_224_INT8.tflite \ + --output ./resources_downloaded/img_class/mobilenet_v2_1.0_224_INT8.tflite +curl -L https://github.com/ARM-software/ML-zoo/raw/e0aa361b03c738047b9147d1a50e3f2dcb13dbcb/models/image_classification/mobilenet_v2_1.0_224/tflite_int8/testing_input/tfl.quantize/0.npy \ --output ./resources_downloaded/img_class/ifm0.npy -curl -L https://github.com/ARM-software/ML-zoo/raw/68b5fbc77ed28e67b2efc915997ea4477c1d9d5b/models/image_classification/mobilenet_v2_1.0_224/tflite_uint8/testing_output/output/0.npy \ +curl -L https://github.com/ARM-software/ML-zoo/raw/e0aa361b03c738047b9147d1a50e3f2dcb13dbcb/models/image_classification/mobilenet_v2_1.0_224/tflite_int8/testing_output/MobilenetV2/Predictions/Reshape_11/0.npy \ --output ./resources_downloaded/img_class/ofm0.npy curl -L https://github.com/ARM-software/ML-zoo/raw/68b5fbc77ed28e67b2efc915997ea4477c1d9d5b/models/keyword_spotting/ds_cnn_large/tflite_clustered_int8/ds_cnn_clustered_int8.tflite \ --output ./resources_downloaded/kws/ds_cnn_clustered_int8.tflite @@ -137,13 +137,13 @@ mv resources_downloaded/kws_asr/ds_cnn_clustered_int8_vela.tflite resources_down --output-dir=resources_downloaded/inference_runner mv resources_downloaded/inference_runner/dnn_s_quantized_vela.tflite resources_downloaded/inference_runner/dnn_s_quantized_vela_H128.tflite -. resources_downloaded/env/bin/activate && vela resources_downloaded/img_class/mobilenet_v2_1.0_224_quantized_1_default_1.tflite \ +. resources_downloaded/env/bin/activate && vela resources_downloaded/img_class/mobilenet_v2_1.0_224_INT8.tflite \ --accelerator-config=ethos-u55-128 \ --optimise Performance --config scripts/vela/default_vela.ini \ --memory-mode=Shared_Sram \ --system-config=Ethos_U55_High_End_Embedded \ --output-dir=resources_downloaded/img_class -mv resources_downloaded/img_class/mobilenet_v2_1.0_224_quantized_1_default_1_vela.tflite resources_downloaded/img_class/mobilenet_v2_1.0_224_quantized_1_default_1_vela_H128.tflite +mv resources_downloaded/img_class/mobilenet_v2_1.0_224_INT8.tflite_vela.tflite resources_downloaded/img_class/mobilenet_v2_1.0_224_INT8.tflite_vela_H128.tflite . resources_downloaded/env/bin/activate && vela resources_downloaded/asr/wav2letter_int8.tflite \ --accelerator-config=ethos-u55-128 \ diff --git a/docs/use_cases/img_class.md b/docs/use_cases/img_class.md index b3544de..ae74d8a 100644 --- a/docs/use_cases/img_class.md +++ b/docs/use_cases/img_class.md @@ -319,7 +319,6 @@ What the preceding choices do: 4. Show NN model info: Prints information about the model data type, input, and output, tensor sizes: ```log - INFO - uTFL version: 2.5.0 INFO - Model info: INFO - Model INPUT tensors: INFO - tensor type is UINT8 @@ -329,19 +328,20 @@ What the preceding choices do: INFO - 2: 224 INFO - 3: 3 INFO - Quant dimension: 0 - INFO - Scale[0] = 0.007812 - INFO - ZeroPoint[0] = 128 + INFO - Scale[0] = 0.007843 + INFO - ZeroPoint[0] = -1 INFO - Model OUTPUT tensors: - INFO - tensor type is UINT8 + INFO - tensor type is INT8 INFO - tensor occupies 1001 bytes with dimensions INFO - 0: 1 INFO - 1: 1001 INFO - Quant dimension: 0 - INFO - Scale[0] = 0.098893 - INFO - ZeroPoint[0] = 58 - INFO - Activation buffer (a.k.a tensor arena) size used: 521760 + INFO - Scale[0] = 0.03906 + INFO - ZeroPoint[0] = -128 + INFO - Activation buffer (a.k.a tensor arena) size used: 1510012 INFO - Number of operators: 1 INFO - Operator 0: ethos-u + ``` 5. List Images: Prints a list of pair image indexes. The original filenames are embedded in the application, like so: @@ -364,18 +364,18 @@ The following example illustrates an application output for classification: INFO - Running inference on image 0 => cat.bmp INFO - Final results: INFO - Total number of inferences: 1 -INFO - 0) 282 (14.636096) -> tabby, tabby cat -INFO - 1) 286 (14.537203) -> Egyptian cat -INFO - 2) 283 (12.757138) -> tiger cat -INFO - 3) 458 (7.021370) -> bow tie, bow-tie, bowtie -INFO - 4) 288 (7.021370) -> lynx, catamount +INFO - 0) 282 (0.753906) -> tabby, tabby cat +INFO - 1) 286 (0.148438) -> Egyptian cat +INFO - 2) 283 (0.062500) -> tiger cat +INFO - 3) 458 (0.003906) -> bow tie, bow-tie, bowtie +INFO - 4) 288 (0.003906) -> lynx, catamount INFO - Profile for Inference: -INFO - NPU AXI0_RD_DATA_BEAT_RECEIVED beats: 2489726 -INFO - NPU AXI0_WR_DATA_BEAT_WRITTEN beats: 1098726 -INFO - NPU AXI1_RD_DATA_BEAT_RECEIVED beats: 471129 -INFO - NPU ACTIVE cycles: 7489258 -INFO - NPU IDLE cycles: 914 -INFO - NPU TOTAL cycles: 7490172 +INFO - NPU AXI0_RD_DATA_BEAT_RECEIVED beats: 2468259 +INFO - NPU AXI0_WR_DATA_BEAT_WRITTEN beats: 1151319 +INFO - NPU AXI1_RD_DATA_BEAT_RECEIVED beats: 432351 +INFO - NPU ACTIVE cycles: 7345741 +INFO - NPU IDLE cycles: 431 +INFO - NPU TOTAL cycles: 7346172 ``` It can take several minutes to complete one inference run. The average time is around 2-3 minutes. @@ -387,18 +387,18 @@ The profiling section of the log shows that for this inference: - *Ethos-U* PMU report: - - 7,490,172 total cycle: The number of NPU cycles. + - 7,346,172 total cycle: The number of NPU cycles. - - 7,489,258 active cycles: The number of NPU cycles that were used for computation. + - 7,345,741 active cycles: The number of NPU cycles that were used for computation. - - 914 idle cycles: The number of cycles for which the NPU was idle. + - 413 idle cycles: The number of cycles for which the NPU was idle. - - 2,489,726 AXI0 read beats: The number of AXI beats with read transactions from AXI0 bus. AXI0 is the bus where the + - 2,468,259 AXI0 read beats: The number of AXI beats with read transactions from AXI0 bus. AXI0 is the bus where the *Ethos-U* NPU reads and writes to the computation buffers, activation buf, or tensor arenas. - - 1,098,726 AXI0 write beats: The number of AXI beats with write transactions to AXI0 bus. + - 1,151,319 AXI0 write beats: The number of AXI beats with write transactions to AXI0 bus. - - 471,129 AXI1 read beats: The number of AXI beats with read transactions from AXI1 bus. AXI1 is the bus where the + - 432,351 AXI1 read beats: The number of AXI beats with read transactions from AXI1 bus. AXI1 is the bus where the *Ethos-U* NPU reads the model. So, read-only. - For FPGA platforms, a CPU cycle count can also be enabled. However, do not use cycle counters for FVP, as the CPU |