yzhliu removed a comment on issue #4259: [DEV][DRAFT] TVM v0.6 Release candidate URL: https://github.com/apache/incubator-tvm/issues/4259#issuecomment-552266647 # Build and Test * Increate the robuteness of CI test ([#2841](https://github.com/dmlc/tvm/pull/2841), [#2798](https://github.com/dmlc/tvm/pull/2798), [#2793](https://github.com/dmlc/tvm/pull/2793), [#2788](https://github.com/dmlc/tvm/pull/2788), [#2781](https://github.com/dmlc/tvm/pull/2781), [#2727](https://github.com/dmlc/tvm/pull/2727), [#2710](https://github.com/dmlc/tvm/pull/2710), [#2711](https://github.com/dmlc/tvm/pull/2711), [#2923](https://github.com/dmlc/tvm/pull/2923)) * Improve conda build ([#2742](https://github.com/dmlc/tvm/pull/2742)) * Add caffe2 nnvm frontend to CI (#3018) * Use bridge network and expose port on macOS when launch docker image (#3086） * Run DarkNet tests (#2673) * Add file type check (#3116) * Always run cpptest during build to ensure library correctness ([#3147](https://github.com/dmlc/tvm/pull/3147)) * Handle more file types in ASF header ([#3235](https://github.com/dmlc/tvm/pull/3235)) * Add `test_forward_ssd_mobilenet_v1` to tflite/test_forward ([#3350](https://github.com/dmlc/tvm/pull/3350)) * Add Azure build pipeline ([#3458](https://github.com/dmlc/tvm/pull/3458), [#3459](https://github.com/dmlc/tvm/pull/3459)) * Update ci-gpu to v0.52 ([#3374](https://github.com/dmlc/tvm/pull/3374)) * Enable more visible symbols by default ([#3365](https://github.com/dmlc/tvm/pull/3365)) * Separate out legacy as a stage in CI ([#3337](https://github.com/dmlc/tvm/pull/3337)) * Simplify build script, remove python 2 support ([#3419](https://github.com/dmlc/tvm/pull/3419)) * Ignore rust cargo lock files in rat ([#3314](https://github.com/dmlc/tvm/pull/3314)) * Improve CUDA Conda package build ([#3281](https://github.com/dmlc/tvm/pull/3281)) * Update CMakeLists.txt to be more flexible to find the third parties libraries ([#3354](https://github.com/dmlc/tvm/pull/3354)) * Docker update conda package ([#3344](https://github.com/dmlc/tvm/pull/3344)), requests and pillow ([#3495](https://github.com/dmlc/tvm/pull/3495)), Android demo ([#3499](https://github.com/dmlc/tvm/pull/3499)), rat install ([#3527](https://github.com/dmlc/tvm/pull/3527)), ARM support ([#3546](https://github.com/dmlc/tvm/pull/3546)), LLVM ([#3590](https://github.com/dmlc/tvm/pull/3590)) * Relay-to-Python testing ([#3156](https://github.com/dmlc/tvm/pull/3156)) * Code refactoring/remove ([#3523](https://github.com/dmlc/tvm/pull/3523), [#3667](https://github.com/dmlc/tvm/pull/3667)) * Zero-rank testing ([#3612](https://github.com/dmlc/tvm/pull/3612)) * CMake compilation ([#3611](https://github.com/dmlc/tvm/pull/3611), [#3650](https://github.com/dmlc/tvm/pull/3650), google test [#3628](https://github.com/dmlc/tvm/pull/3628)) * Standalone wheel build for TOPI ([#3657](https://github.com/dmlc/tvm/pull/3657)) * Fixing performance issues in PassUpDomain when fusing and splitting axes ([#3073](https://github.com/dmlc/tvm/pull/3073)) * conda recipe ([#3791](https://github.com/dmlc/tvm/pull/3791)) * Allow users to specify download directory ([#3803](https://github.com/dmlc/tvm/pull/3803)) * Update docs for installation for CUDA ([#3832](https://github.com/dmlc/tvm/pull/3832)) * Update hybrid_script.rst ([#3799](https://github.com/dmlc/tvm/pull/3799)) * Acknowledge Halide attributions ([#3824](https://github.com/dmlc/tvm/pull/3824)) * Add psutil dependency ([#3780](https://github.com/dmlc/tvm/pull/3780)) * Temporary disable rust test ([#3809](https://github.com/dmlc/tvm/pull/3809)) * Solve occasional CI issue when pad value is all 0 ([#3801](https://github.com/dmlc/tvm/pull/3801)) * Towards TSIM CI testing ([#3704](https://github.com/dmlc/tvm/pull/3704)) * Use pip3 for python3 ([#3742](https://github.com/dmlc/tvm/pull/3742)) * Update docker image `ci_cpu,i386` to include verilator ([#3738](https://github.com/dmlc/tvm/pull/3738)) * Remove sccache from Rust install ([#3728](https://github.com/dmlc/tvm/pull/3728)) * Update dmlc-core to the latest commit ([#3716](https://github.com/dmlc/tvm/pull/3716)) * Update GPU docker ([#3709](https://github.com/dmlc/tvm/pull/3709)) * Add an option to build with -pthread ([#3671](https://github.com/dmlc/tvm/pull/3671)) * Add DGL to `{ci_gpu, demo_cpu, demo_gpu}` docker images ([#3692](https://github.com/dmlc/tvm/pull/3692)) * Use pytest instead of nosetest ([#3524](https://github.com/dmlc/tvm/pull/3524)) * Enable NHWC of `relay.testing.mobilenet` ([#3886](https://github.com/dmlc/tvm/pull/3886)) * Add .hsaco save/load for tesnor_expr Tutorial ([#3852](https://github.com/dmlc/tvm/pull/3852)) * Support LLVM trunk ([#3907](https://github.com/dmlc/tvm/pull/3907)) * Remove GTest cmake flag from install docs ([#3953](https://github.com/dmlc/tvm/pull/3953)) * Allow `USE_LLVM` to take extra arguments ([#3954](https://github.com/dmlc/tvm/pull/3954)) * [CI] Pin NNPack pthreadtools version ([#4152](https://github.com/dmlc/tvm/pull/4152)) * [TOPI] Fix flaky testcase for check round ([#4211](https://github.com/dmlc/tvm/pull/4211)) * [CI] Move gpu docker binary to cuda10 ([#4229](https://github.com/dmlc/tvm/pull/4229)) * [CI] use llvm9 for the gpu tests ([#4224](https://github.com/dmlc/tvm/pull/4224)) * [CI] Update GPU docker to cuda10 ([#4228](https://github.com/dmlc/tvm/pull/4228)) * [Relay] Install Relay Prelude program in package install ([#4227](https://github.com/dmlc/tvm/pull/4227)) * [relay] use time_evaluator for measurement ([#4191](https://github.com/dmlc/tvm/pull/4191)) * [Relay] Improve build error when no lowered funcs are produced ([#4132](https://github.com/dmlc/tvm/pull/4132)) * [llvm] switch to use Align for llvm trunk ([#4051](https://github.com/dmlc/tvm/pull/4051)) * [CUDA] Update have_int8 condition to run on compute capability 7.x devices ([#4214](https://github.com/dmlc/tvm/pull/4214)) * [DOCKER] Pin torchvision==0.4.1 ([#4140](https://github.com/dmlc/tvm/pull/4140)) * [DOCKER] torch install depends on future package ([#4098](https://github.com/dmlc/tvm/pull/4098)) * [CodeGen] Disable -mfloat-abi hard option for LLVM < 6.0 ([#4071](https://github.com/dmlc/tvm/pull/4071)) * Add a python how to example of deploying tvm module with tvm runtime only ([#4094](https://github.com/dmlc/tvm/pull/4094)) * Hide symbols from dependent libraries if HIDE_PRIVATE_SYMBOLS is ON. ([#4041](https://github.com/dmlc/tvm/pull/4041)) * [BUILD] Disable utvm standalone runtime by default ([#4240](https://github.com/dmlc/tvm/pull/4240)) * Fix TSIM compile error in Linux (add missing -fPIC flag) ([#3876](https://github.com/dmlc/tvm/pull/3876)) * Add scalafmt and format existing scala codebase ([#3880](https://github.com/dmlc/tvm/pull/3880)) * Update TFLite wheel version to 1.13.1 ([#3435](https://github.com/dmlc/tvm/pull/3435)) # Bug Fixes * [RELAY] Fix `get_int_tuple`. (#2691) * [ARITH] Select support for integer set analysis. (#2687) * [Relay] Fix error in ANF (too agressively inline atomic expression and create free variable). (#2665) * [Hybrid Script] Fix name conflict and attached scope problem. (#2649) * [Relay] Fix ANF for reference and pattern matching. (#2637) * [Relay] Fix fusion bug when call symbol that is not an operator. (#2630) * Fix missing <sstream> header file. (#2629) * [Relay]Fix the bug in heterogeneous annotation which mistakenly steps into the fused op. (#2622) * [AutoTVM] Fix incorrect localhost usage in RPC mode. (#2619) * [NNVM] Fix incorrectly getting layout attribute as a tuple. (#2610) * [Relay] Fix mutating IF expression. (#2601) * [Tutorial] Fix downloaded file path. (#2590) * [Storage] Fix int32 overflow bug when input is big. (#2580) * [NNVM] Fix non-identity problem for FInplaceIdentity. (#2572) * [Golang] Fix compilation error. (#2558) * [Tensor Expression] Fix missing reduction init predicates. (#2495) * [Relay] Fix missing argument for NCHWc in Relay. (#2627) * [TOPI] Fix Nms_ir data race. (#2600) * Fix compute_inline with multiple outputs ([#2934](https://github.com/dmlc/tvm/pull/2934)) * [TEXPR][PASS] Fix thread all reduce to avoid write after read hazzard ([#2937](https://github.com/dmlc/tvm/pull/2937)) * [FRONTEND][TENSORFLOW] bug fix for tensorflow official slim models. ([#2864](https://github.com/dmlc/tvm/pull/2864)) * [FRONTEND][ONNX] Some bug fixes and Shape operator fixed for relay. ([#2850](https://github.com/dmlc/tvm/pull/2850)) * Turn on USE_SORT by default ([#2916](https://github.com/dmlc/tvm/pull/2916)) * [DOCKER] Upgrade ci-cpu to latest v0.50 ([#2901](https://github.com/dmlc/tvm/pull/2901)) * [TESTS] Import script robustness (set -u) ([#2896](https://github.com/dmlc/tvm/pull/2896)) * [Relay] Fix name of bias in testing.mlp ([#2892](https://github.com/dmlc/tvm/pull/2892)) * [TESTS] Improve script robustness ([#2893](https://github.com/dmlc/tvm/pull/2893)) * Add dense schedules to `__init__` for cpu ([#2855](https://github.com/dmlc/tvm/pull/2855)) * [Apps] [howto_deploy] fix cxx-flags order and build directory ([#2888](https://github.com/dmlc/tvm/pull/2888)) * [Relay] Add TVM_DLL for ANF/GNF conversion [#2883](https://github.com/dmlc/tvm/pull/2883) * [Relay] Fix Relay ARM CPU depthwise spatial pack schedule alter op layout issue. ([#2861](https://github.com/dmlc/tvm/pull/2861)) * Fix setting up hints for getaddrinfo ([#2872](https://github.com/dmlc/tvm/pull/2872)) * Add missing sgx includes ([#2878](https://github.com/dmlc/tvm/pull/2878)) * Fix error reporting for missing axis ([#2835](https://github.com/dmlc/tvm/pull/2835)) * Fix an OrderDict initilization bug. ([#2862](https://github.com/dmlc/tvm/pull/2862)) * Fix Xcode 10 metal compile error ([#2836](https://github.com/dmlc/tvm/pull/2836)) * tvmrpc: Fix includes ([#2825](https://github.com/dmlc/tvm/pull/2825)) * Fix `init_proj.py`: Team ID expected ([#2824](https://github.com/dmlc/tvm/pull/2824)) * [DOCKER] Fix git clone failure. ([#2816](https://github.com/dmlc/tvm/pull/2816)) * upgrade java style-check due to CVE-2019-9658 ([#2817](https://github.com/dmlc/tvm/pull/2817)) * [Relay][Quantization] Fix duplicated simulated quantization ([#2803](https://github.com/dmlc/tvm/pull/2803)) * [Bugfix] Repeat and tile bug fixed, relay tests added ([#2804](https://github.com/dmlc/tvm/pull/2804)) * Fix caffe2 relay frontend ([#2733](https://github.com/dmlc/tvm/pull/2733)) * Fix a bug in nnvm to relay converter. ([#2756](https://github.com/dmlc/tvm/pull/2756)) * Ensure loop count is a constant before trying to unroll. ([#2797](https://github.com/dmlc/tvm/pull/2797)) * xcode.py: Decode bytes before output [#2833](https://github.com/dmlc/tvm/pull/2833) * [WIN] Fix a bug in `find_llvm` when specify llvm-config ([#2758](https://github.com/dmlc/tvm/pull/2758)) * [DLPACK] fix flaky ctypes support ([#2759](https://github.com/dmlc/tvm/pull/2759)) * [Bugfix][Relay][Frontend] Fix bug in mxnet converter for slick_like ([#2744](https://github.com/dmlc/tvm/pull/2744)) * [DOCS] Fix tutorial ([#2724](https://github.com/dmlc/tvm/pull/2724)) * [TOPI][Relay] Fix default `out_dtype` for `conv2d_NCHWc` and Relay ([#2702](https://github.com/dmlc/tvm/pull/2702)) * [Relay] fix checkwellform ([#2705](https://github.com/dmlc/tvm/pull/2705)) fix prelu, now can use on 2d input and add one test ([#2875](https://github.com/dmlc/tvm/pull/2875)) * [CODEGEN][OPENCL] Fix compile error about ternary expression. ([#2821](https://github.com/dmlc/tvm/pull/2821)) * Fix Placeholder issue (#2834) * Fix makedirs() condition in contrib (#2942) * Add missing #!/bin/bash directive (#2951) * Bilinear resize bug fix from PR #2777 (#2857) * Fix bias_add default axis (#2829) * Remove empty ty.rs (#2958) * fix undefined reference to dlopen, etc (#2957) * Removed deprecated `std::unary_function` (#2962) * Add output format to ndk build func (#2999) * Fix java checkstyle version (#2998) * Fix relay invariant error message (#3011) * Fix for caffe2 nnvm frontend (#2996) * Fix rust resnet example (#3000) * Fix x||!x for comparisons in rewrite simplifier (#3029) * Fix BatchMatMulRel typerelation (#3032) * Update dmlc-core, fix default ctors of NodeEntry (#3017) * Fix Fuse (#3035) * Fix PostOrderVisit signature (#3048) * Fix winograd nnpack fp16 (#3046) * Fix some typos (#3063, #3112) * Fix group_conv2d unit test (#3113) * Fix bug in ONNX importer (#3084) * Fixing a doc nit (#3123) * Fix type code error for StringImm (#3050) * Fix bug of wrongly generated device_map (#2990) * use unordered_map instead of map in ANF (#3024) * Fix PRelu layout in Relay ([#3013](https://github.com/dmlc/tvm/pull/3013)) * Minor addition to graph runtime debug ([#3129](https://github.com/dmlc/tvm/pull/3129)) * Fix mali conv2d performance regression ([#3131](https://github.com/dmlc/tvm/pull/3131)) * Fix dense autotvm template registration in ROCm ([#3136](https://github.com/dmlc/tvm/pull/3136)) * Fix `conv2d_transpose` ([#3138](https://github.com/dmlc/tvm/pull/3138)) * Fix python lint warnings ([#3145](https://github.com/dmlc/tvm/pull/3145)) * Some fixes for golang latest version compiler #3119 ([#3182](https://github.com/dmlc/tvm/pull/3182)) * Add more syncs to fix flaky test caused by `get_valid_counts` ([#3151](https://github.com/dmlc/tvm/pull/3151)) * Fix AlterLayout Pass ([#3155](https://github.com/dmlc/tvm/pull/3155)) * Fix a multithreaded bug in llvm LazyInitJIT ([#3158](https://github.com/dmlc/tvm/pull/3158)) * Fix a tensorflow test bug. ([#3165](https://github.com/dmlc/tvm/pull/3165)) * Fix concat for ARM ([#3061](https://github.com/dmlc/tvm/pull/3061)) * Handle vectorize for LE statement ([#3137](https://github.com/dmlc/tvm/pull/3137)) * Raise exception `group_conv2d_nchw` not supported ([#3195](https://github.com/dmlc/tvm/pull/3195)) * Quick fix of VTA FPGA Toolchain Installation documentation ([#3196](https://github.com/dmlc/tvm/pull/3196)) * Check file exists before removing it ([#3178](https://github.com/dmlc/tvm/pull/3178)) * Fix a bug of flatten in ONNX to Relay converter ([#3180](https://github.com/dmlc/tvm/pull/3180)) * Fix converter where initializers were not registered as nodes ([#3143](https://github.com/dmlc/tvm/pull/3143)) * Fix bug in cast to bool ([#3207](https://github.com/dmlc/tvm/pull/3207)) * Hotfix `build_module` creation ([#3198](https://github.com/dmlc/tvm/pull/3198)) * Fix sort changing original input data issue ([#3212](https://github.com/dmlc/tvm/pull/3212)) * Fix bug in vta runtime DepPop function ([#3208](https://github.com/dmlc/tvm/pull/3208)) * Fix resize nearest with fractional scaling ([#3244](https://github.com/dmlc/tvm/pull/3244)) * Fix `vta_conv2d` crash issue after change `vta_config.json` ([#3213](https://github.com/dmlc/tvm/pull/3213)) * Fix a memory leak in OpManager ([#3263](https://github.com/dmlc/tvm/pull/3263)) * PkgConfig cause crash in PYNQ board due to link library ([#3257](https://github.com/dmlc/tvm/pull/3257)) * Fix Error messages in tflite.py ([#3320](https://github.com/dmlc/tvm/pull/3320)) * Fix typos in docs and comments ([#3309](https://github.com/dmlc/tvm/pull/3309), [#3376](https://github.com/dmlc/tvm/pull/3376)) * Bugfix min/max const canonicalize rule ([#3386](https://github.com/dmlc/tvm/pull/3386)) * Return module from frontend for autotvm ([#3401](https://github.com/dmlc/tvm/pull/3401)) * Fix constant and reshape in ONNX ([#3387](https://github.com/dmlc/tvm/pull/3387)) * Default verilator location fix ([#3324](https://github.com/dmlc/tvm/pull/3324)) * Fix autodiff for conditional expression ([#3453](https://github.com/dmlc/tvm/pull/3453)) * Gramatical improvements to `tensor_expr_get_started` ([#3330](https://github.com/dmlc/tvm/pull/3330)) * Fix AutoTVM data structure bug ([#3462](https://github.com/dmlc/tvm/pull/3462)) * Fix MXNet RNN without providing state initialization as input ([#3326](https://github.com/dmlc/tvm/pull/3326)) * Fix flaky test on topk and quantize pass ([#3362](https://github.com/dmlc/tvm/pull/3362)) * Add VTA PYNQ metal_test bitstream program logic and fix compilation issue. ([#3400](https://github.com/dmlc/tvm/pull/3400)) * Fix VTA function Vivado Compile Error. ([#3375](https://github.com/dmlc/tvm/pull/3375)) * Fix VTA DRAM functionality issue. ([#3278](https://github.com/dmlc/tvm/pull/3278)) * Fix reshape precompute and type error in ONNX frontend ([#3230](https://github.com/dmlc/tvm/pull/3230)) * Fix interpreter argument conversion for tuples. ([#3349](https://github.com/dmlc/tvm/pull/3349)) * Fix code generation for packed functions + tuples in VM ([#3287](https://github.com/dmlc/tvm/pull/3287)) * Fix memory leak in Relay interpreter ([#3448](https://github.com/dmlc/tvm/pull/3448)) * Fix x86 depthwise conv2d `alter_op_layout` ([#3264](https://github.com/dmlc/tvm/pull/3264)) * Create closure object for GlobalVar ([#3411](https://github.com/dmlc/tvm/pull/3411)) * Fix getting global var in prelude ([#3405](https://github.com/dmlc/tvm/pull/3405)) * Fix rfactor bugs which related to predicate and loop partition ([#3382](https://github.com/dmlc/tvm/pull/3382), [#3444](https://github.com/dmlc/tvm/pull/3444)) * Fix the bug in AutoTVM where SimulatedAnnealingOptimizer sometimes finds useless candidate ([#3413](https://github.com/dmlc/tvm/pull/3413)) * Fix name conflict in PartialEval ([#3402](https://github.com/dmlc/tvm/pull/3402)) * Fix int bound analysis bug for modular ([#3288](https://github.com/dmlc/tvm/pull/3288)) * Check arg positiveness for modular rules ([#3279](https://github.com/dmlc/tvm/pull/3279)) * Fixes failure of `sum` and `all` on `axis=0` ([#3422](https://github.com/dmlc/tvm/pull/3422)) * Fix package path in tflite test ([#3427](https://github.com/dmlc/tvm/pull/3427)) * Fix Windows build ([#3429](https://github.com/dmlc/tvm/pull/3429)) * Fix `LSTMBlockCell` in Tensorflow frontend ([#3410](https://github.com/dmlc/tvm/pull/3410)) * TF fix where output index is ignored ([#3622](https://github.com/dmlc/tvm/pull/3622)) * Runtime fix for custom datatypes ([#3471](https://github.com/dmlc/tvm/pull/3471)) * Relay build module warnings ([#3452](https://github.com/dmlc/tvm/pull/3452)) * Relay partial evaluator ([#3482](https://github.com/dmlc/tvm/pull/3482)) * Pynq AutoTVM tracker ([#3497](https://github.com/dmlc/tvm/pull/3497), [#3578](https://github.com/dmlc/tvm/pull/3578)) * A normal form test ([#3525](https://github.com/dmlc/tvm/pull/3525)) * Lint issue ([#3519](https://github.com/dmlc/tvm/pull/3519), [#3615](https://github.com/dmlc/tvm/pull/3615) ) * Any shape testing ([#3528](https://github.com/dmlc/tvm/pull/3528)) * Android posix_memalign ([#3532](https://github.com/dmlc/tvm/pull/3532)) * Quantization add_rewrite and UnifyDTypeScale ([#3534](https://github.com/dmlc/tvm/pull/3534)) * Bound inference fix ([#3526](https://github.com/dmlc/tvm/pull/3526)) * Tensorflow NCHW data format ([#3514](https://github.com/dmlc/tvm/pull/3514)) * First order gradient ([#3550](https://github.com/dmlc/tvm/pull/3550)) * JS load module example ([#3556](https://github.com/dmlc/tvm/pull/3556)) * Build error ([#3552](https://github.com/dmlc/tvm/pull/3552)) * Relay VM debug statements ([#3565](https://github.com/dmlc/tvm/pull/3565)) * C++ lambda expr ([#3570](https://github.com/dmlc/tvm/pull/3570)) * Handling of tempdir if subprocess is killed ([#3574](https://github.com/dmlc/tvm/pull/3574)) * Remove tabs in Chisel source ([#3603](https://github.com/dmlc/tvm/pull/3603)) * Relay VM DataTypeObject ([#3604](https://github.com/dmlc/tvm/pull/3604)) * Removing prints ([#3616](https://github.com/dmlc/tvm/pull/3616)) * Average Pool2D Bug ([#3607](https://github.com/dmlc/tvm/pull/3607)) * Missing header in cuda_device_api.cc ([#3621](https://github.com/dmlc/tvm/pull/3621)) * Tensorflow frontend fix where output_shape is None ([#3632](https://github.com/dmlc/tvm/pull/3632)) * Winograd accuracy fix ([#3644](https://github.com/dmlc/tvm/pull/3644)) * Fix comment ([#3646](https://github.com/dmlc/tvm/pull/3646)) * Zero-input op fix for recursive traversals ([#3623](https://github.com/dmlc/tvm/pull/3623)) * Python 3.5 compatibility ([#3675](https://github.com/dmlc/tvm/pull/3675)) * Fix infinite recursive `device_api.ext_dev` call in VTA. ([#3843](https://github.com/dmlc/tvm/pull/3843)) * Fix depth_mult for TensorFlow frontend ([#3676](https://github.com/dmlc/tvm/pull/3676)) * Fix database APIs for AutoTVM ([#3821](https://github.com/dmlc/tvm/pull/3821)) * Fix axis of softmax in Keras ([#3834](https://github.com/dmlc/tvm/pull/3834)) * Fix VTA TensorLoad module ([#3841](https://github.com/dmlc/tvm/pull/3841)) * Fix inconsistent python/cpp API behavior for `if_then_else`, power ([#3829](https://github.com/dmlc/tvm/pull/3829)) * Fix code comment of operators in ONNX frontend ([#3830](https://github.com/dmlc/tvm/pull/3830)) * Added repo for llvm-9 to fix missing dependency issue ([#3826](https://github.com/dmlc/tvm/pull/3826)) * Fix typo in Relay text parser ([#3785](https://github.com/dmlc/tvm/pull/3785)) * Fix tvm const warnings ([#3817](https://github.com/dmlc/tvm/pull/3817)) * Add gfx906 bc ([#3808](https://github.com/dmlc/tvm/pull/3808)) * Fixed onnx test failures when run on a cpu backend ([#3764](https://github.com/dmlc/tvm/pull/3764)) * Fix ArgBinder assert order ([#3794](https://github.com/dmlc/tvm/pull/3794)) * Fix for NoneType Target for quantization ([#3792](https://github.com/dmlc/tvm/pull/3792)) * Fix out-of-date quantization realize ([#3790](https://github.com/dmlc/tvm/pull/3790)) * Fix Qnn concatenate InferType ([#3779](https://github.com/dmlc/tvm/pull/3779)) * Fix dense tuning ([#3768](https://github.com/dmlc/tvm/pull/3768)) * Fix `visit_pattern` in ExprMutator ([#3769](https://github.com/dmlc/tvm/pull/3769)) * Fix Chisel Scala style ([#3765](https://github.com/dmlc/tvm/pull/3765)) * Fix some pass docs ([#3767](https://github.com/dmlc/tvm/pull/3767)) * Fix mistype in rpc tutorial ([#3763](https://github.com/dmlc/tvm/pull/3763)) * Fix tvm.scan follow by tvm.compute segfault ([#3723](https://github.com/dmlc/tvm/pull/3723)) * Fix the potential index overflow in where operator ([#3751](https://github.com/dmlc/tvm/pull/3751)) * Revert `compile_cmd` kwarg name change ([#3746](https://github.com/dmlc/tvm/pull/3746)) * Update tophub ([#3752](https://github.com/dmlc/tvm/pull/3752)) * Fix typo in `ir_pass.h` ([#3741](https://github.com/dmlc/tvm/pull/3741)) * Bug fix for VME Shell ([#3737](https://github.com/dmlc/tvm/pull/3737)) * Fix missing apt https transport support ([#3735](https://github.com/dmlc/tvm/pull/3735)) * Take zero extent loops as NoOp and remove it ([#3724](https://github.com/dmlc/tvm/pull/3724)) * Fix mxnet converter for hybridblock and add `div_sqrt_dim` ([#3701](https://github.com/dmlc/tvm/pull/3701)) * Fix partial eval unit test name ([#3719](https://github.com/dmlc/tvm/pull/3719)) * Fix conv2d schedule code ([#3648](https://github.com/dmlc/tvm/issues/3648), [#3717](https://github.com/dmlc/tvm/pull/3717)) * Remove thread related headers ([#3713](https://github.com/dmlc/tvm/pull/3713)) * Fix FunctionPass ([#3712](https://github.com/dmlc/tvm/pull/3712)) * Export tvm::relay::OpRegistry::OpRegistry ([#3711](https://github.com/dmlc/tvm/pull/3711)) * Fix Metal reinterpret ([#3706](https://github.com/dmlc/tvm/pull/3706)) * Fix `gather_nd` in Relay ([#3442](https://github.com/dmlc/tvm/pull/3442)) * Fix error in partial evaluator ([#3693](https://github.com/dmlc/tvm/pull/3693)) * Align the naming rule for OpAttributeUnImplemented ([#3695](https://github.com/dmlc/tvm/pull/3695)) * Enable the sparse schedule ([#3651](https://github.com/dmlc/tvm/pull/3651)) * Fix typo names in Caffe2 frontend ([#3685](https://github.com/dmlc/tvm/pull/3685)) * Make tests multi-process friendly. ([#3683](https://github.com/dmlc/tvm/pull/3683)) * Fix typo in README.md ([#3684](https://github.com/dmlc/tvm/pull/3684)) * Fix doc rendering ([#3897](https://github.com/dmlc/tvm/pull/3897)) * Add test script starter command to document ([#3993](https://github.com/dmlc/tvm/pull/3993)) * Add type solver unit tests for unifying quantified funcs ([#3947](https://github.com/dmlc/tvm/pull/3947)) * Change Vivado install instructions to version 2018.3 ([#4003](https://github.com/dmlc/tvm/pull/4003)) * Add a link to the defining network description of auto-tuning tutorial ([#4023](https://github.com/dmlc/tvm/pull/4023)) * Additional MXNet Convolution and Deconvolution tests ([#4026](https://github.com/dmlc/tvm/pull/4026)) * Adding support to check if an attribute is present or not without having to get the value ([#3957](https://github.com/dmlc/tvm/pull/3957)) * Fix parser for cast. ([#3873](https://github.com/dmlc/tvm/pull/3873)) * Fix operator fusion for multiple output ([#3871](https://github.com/dmlc/tvm/pull/3871)) * Remove extern C warpper for cuBLAS ([#3877](https://github.com/dmlc/tvm/pull/3877)) * Fix int32 range overflow by using int64 ([#3870](https://github.com/dmlc/tvm/pull/3870)) * Remove duplicate resize ([#3902](https://github.com/dmlc/tvm/pull/3902)) * Fix blas cmake for mac os ([#3898](https://github.com/dmlc/tvm/pull/3898)) * Add another MKL name alias for MKL installed through pypi ([#3853](https://github.com/dmlc/tvm/pull/3853)) * Numpy compatible dtype inference for `tvm.convert` and `tvm.const` ([#3861](https://github.com/dmlc/tvm/pull/3861)) * Remove incorrect check for LLVM in C codegen test ([#3921](https://github.com/dmlc/tvm/pull/3921)) * Fix exponential blowup in interpreter ([#3559](https://github.com/dmlc/tvm/pull/3559)) * Fix CUDA int8x4 vectorize ([#3928](https://github.com/dmlc/tvm/pull/3928)) * Make buffer auto broadcast independent to the order of input args ([#3956](https://github.com/dmlc/tvm/pull/3956)) * Fix benchmark layout in graph tuner ([#3926](https://github.com/dmlc/tvm/pull/3926)) * Fix Android Demo LLVM version ([#3962](https://github.com/dmlc/tvm/pull/3962)) * Cast filepath arguments to string ([#3968](https://github.com/dmlc/tvm/pull/3968)) * Fixes "common" sub crate using nightly and master ([#3965](https://github.com/dmlc/tvm/pull/3965)) * Changes to make tensorize work. These changes also fix the previously broken test. ([#3981](https://github.com/dmlc/tvm/pull/3981)) * Remove FLOP computation when calling 3rd party library ([#4005](https://github.com/dmlc/tvm/pull/4005)) * Use a more intuitive way to limit the #ops in a group ([#4018](https://github.com/dmlc/tvm/pull/4018)) * Add more `pad_mode` support for onnx converter ([#4029](https://github.com/dmlc/tvm/pull/4029)) * Impose a max op limit to the op fusion pass ([#4002](https://github.com/dmlc/tvm/pull/4002)) * Fixes issue with CPP enums ([#4019](https://github.com/dmlc/tvm/pull/4019)) * Int64 shape handling for outputs. ([#4031](https://github.com/dmlc/tvm/pull/4031)) * [PYTHON] Fix installation for generated grammar ([#4223](https://github.com/dmlc/tvm/pull/4223)) * [Bugfix] Fix target host for vm compiler ([#4057](https://github.com/dmlc/tvm/pull/4057)) * [Fix][VM] Fix VM invoke with set_params ([#4079](https://github.com/dmlc/tvm/pull/4079)) * [Fix] Fix a few bugs when dtype is fp16 ([#4088](https://github.com/dmlc/tvm/pull/4088)) * [Relay][Frontend][TF] Fix Size operator ([#4175](https://github.com/dmlc/tvm/pull/4175)) * [cmake][ANTLR] Support setting path to ANTLR jar ([#4176](https://github.com/dmlc/tvm/pull/4176)) * Fix infer type of kernel in dense. ([#4125](https://github.com/dmlc/tvm/pull/4125)) * [Relay] Fix match case in Python-side expr functor ([#4037](https://github.com/dmlc/tvm/pull/4037)) * Split adaptive_pool2d_avg into sum and div ([#4186](https://github.com/dmlc/tvm/pull/4186)) * [AutoTVM] Fix Split Factors when no_tail is off ([#4044](https://github.com/dmlc/tvm/pull/4044)) * Fix extent one for the `post_stmt` in loop partition ([#3734](https://github.com/dmlc/tvm/pull/3734)) * [TOPI] Fix bug in intel graphics auto tune ([#4093](https://github.com/dmlc/tvm/pull/4093)) * [ARITH] Fix lowering of `floormod(x, y) != 0` ([#4127](https://github.com/dmlc/tvm/pull/4127)) * [ARITH] Fix the rule `y < x && x <= y` ([#4220](https://github.com/dmlc/tvm/pull/4220)) * [Bugfix][TF] reset graph after getting tag of savedmodel ([#4055](https://github.com/dmlc/tvm/pull/4055)) * [Fix] Fix the logic of the number of nodes checking in op fusion ([#4074](https://github.com/dmlc/tvm/pull/4074)) * [VTA] hotfix for de10-nano driver ([#4081](https://github.com/dmlc/tvm/pull/4081)) * Fixing tensor not found issue in bitserial operator ([#4095](https://github.com/dmlc/tvm/pull/4095)) * Fix wrong `n_trial` number in autotvm tutorials' progress bar if `n_trial` is larger then config space. ([#4070](https://github.com/dmlc/tvm/pull/4070)) * [PATCH] Fix undefined `__floatdihf` in libtvmruntime.so on aarch64. ([#4119](https://github.com/dmlc/tvm/pull/4119))
---------------------------------------------------------------- This is an automated message from the Apache Git Service. To respond to the message, please log on to GitHub and use the URL above to go to the specific comment. For queries about this service, please contact Infrastructure at: us...@infra.apache.org With regards, Apache Git Services
