-
Notifications
You must be signed in to change notification settings - Fork 741
Commit
This commit does not belong to any branch on this repository, and may belong to a fork outside of the repository.
Add FusedMatmulBiasAddReluDropout [OneEmbedding] (#8222)
* previous version for fused_matmul_bias_add_relu_dropout * add op infer * fix detail * finish forward * support dropout rate list * add forward test * fix bug for output buffer * Configurable alpha params * try to add bit mask logic * Add bitmask first version! * Add row col bitmask logic * support not align4 reludropout * simplify relu dropout ld logic * Add naive relu dropout grad kernel * add simple relu dropout grad kernel * Rename * support relu_dropout bitmask backward * add vectorized optimization * fix tmp buffer * add to amp list * add lazy backward logic * Refine kernel * add indextype dispatch * simplify functor logic * fix cublas fused mlp aux_ld shape bug * Add more relu dropout kernel * add full unittest * fix bug in skip final activation * refine * Remove dump func * fix format * Remove cmake * remove redundant divide * add padded version * fix dropout * oneflow curand * refine * remove redundant kernel * add unroll logic * add unroll and ballot sync * refine format * Remove fast curand * Refine python interface * Add if branch for memset * fix python logic * just for debug * not use matmul bias add grad * add launch 1 block limit * fix unittest * Refine * fix graph backward bug * limit to 11060 * change to use int32_t dtype for cublas aux * Fix jc comment * fix comment * fix convert * fix static_analysis * fix at * fix userops td * fix userops td * fix const ref * fix compile error for bfloat16 * limit to 11060 * fix bug Co-authored-by: Juncheng <liujuncheng1022@gmail.com> Co-authored-by: mergify[bot] <37929162+mergify[bot]@users.noreply.github.com>
- Loading branch information
3 people
authored
Jun 8, 2022
1 parent
e224be4
commit c83fcaa
Showing
19 changed files
with
1,652 additions
and
120 deletions.
There are no files selected for viewing
This file contains bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
205 changes: 205 additions & 0 deletions
205
oneflow/core/autograd/gradient_funcs/fused_matmul_bias_add_relu_dropout.cpp
This file contains bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
| Original file line number | Diff line number | Diff line change |
|---|---|---|
| @@ -0,0 +1,205 @@ | ||
| /* | ||
| Copyright 2020 The OneFlow Authors. All rights reserved. | ||
| Licensed under the Apache License, Version 2.0 (the "License"); | ||
| you may not use this file except in compliance with the License. | ||
| You may obtain a copy of the License at | ||
| http://www.apache.org/licenses/LICENSE-2.0 | ||
| Unless required by applicable law or agreed to in writing, software | ||
| distributed under the License is distributed on an "AS IS" BASIS, | ||
| WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. | ||
| See the License for the specific language governing permissions and | ||
| limitations under the License. | ||
| */ | ||
| #include "oneflow/core/common/error.pb.h" | ||
| #include "oneflow/core/common/just.h" | ||
| #include "oneflow/core/common/maybe.h" | ||
| #include "oneflow/core/framework/op_expr_grad_function.h" | ||
| #include "oneflow/core/framework/op_builder.h" | ||
| #include "oneflow/core/framework/op_expr.h" | ||
| #include "oneflow/core/framework/op_interpreter/op_interpreter_util.h" | ||
| #include "oneflow/core/common/container_util.h" | ||
| #include "oneflow/core/functional/functional.h" | ||
| #include "oneflow/core/functional/functional_api.yaml.h" | ||
| #if CUDA_VERSION >= 11060 | ||
|
|
||
| namespace oneflow { | ||
|
|
||
| namespace one { | ||
|
|
||
| struct FusedMatmulBiasAddReluDropoutCaptureState : public AutoGradCaptureState { | ||
| int32_t weight_num = 0; | ||
| bool skip_final_activation = false; | ||
| bool x_requires_grad = false; | ||
| std::vector<bool> weights_requires_grad; | ||
| std::vector<bool> biases_requires_grad; | ||
| std::vector<float> dropout_rate_list; | ||
| }; | ||
|
|
||
| class FusedMatmulBiasAddReluDropout | ||
| : public OpExprGradFunction<FusedMatmulBiasAddReluDropoutCaptureState> { | ||
| public: | ||
| Maybe<void> Init(const OpExpr& op) override; | ||
| Maybe<void> Capture(FusedMatmulBiasAddReluDropoutCaptureState* ctx, const TensorTuple& inputs, | ||
| const TensorTuple& outputs, const AttrMap& attrs) const override; | ||
| Maybe<void> Apply(const FusedMatmulBiasAddReluDropoutCaptureState* ctx, | ||
| const TensorTuple& out_grads, TensorTuple* in_grads) const override; | ||
|
|
||
| protected: | ||
| AttrMap base_attrs_; | ||
| }; | ||
|
|
||
| Maybe<void> FusedMatmulBiasAddReluDropout::Init(const OpExpr& op) { | ||
| const UserOpExpr* fw_op_expr = dynamic_cast<const UserOpExpr*>(&op); | ||
| CHECK_NOTNULL_OR_RETURN(fw_op_expr); | ||
| base_attrs_ = MakeAttrMapFromUserOpConf(fw_op_expr->proto()); | ||
| return Maybe<void>::Ok(); | ||
| } | ||
|
|
||
| Maybe<void> FusedMatmulBiasAddReluDropout::Capture(FusedMatmulBiasAddReluDropoutCaptureState* ctx, | ||
| const TensorTuple& inputs, | ||
| const TensorTuple& outputs, | ||
| const AttrMap& attrs) const { | ||
| CHECK_OR_RETURN(inputs.size() % 2 == 1) << "Both weight and bias should be passed together. "; | ||
| int32_t weight_num = (inputs.size() - 1) / 2; | ||
| ctx->weight_num = weight_num; | ||
| ctx->x_requires_grad = JUST(VectorAt(inputs, 0))->requires_grad(); | ||
| ctx->weights_requires_grad.resize(weight_num); | ||
| ctx->biases_requires_grad.resize(weight_num); | ||
|
|
||
| for (int32_t i = 0; i < weight_num; i++) { | ||
| ctx->weights_requires_grad.at(i) = inputs.at(i + 1)->requires_grad(); // NOLINT | ||
| ctx->biases_requires_grad.at(i) = inputs.at(i + 1 + weight_num)->requires_grad(); // NOLINT | ||
| } | ||
|
|
||
| ctx->SaveTensorForBackward(JUST(VectorAt(inputs, 0))); // x. idx_sum:1 | ||
| for (int32_t i = 0; i < weight_num; i++) { | ||
| ctx->SaveTensorForBackward(JUST(VectorAt(inputs, i + 1))); // weights. idx_sum:1+w | ||
| } | ||
|
|
||
| ctx->SaveTensorForBackward(JUST(VectorAt(outputs, 0))); // final layers output. idx_sum:2+w | ||
| for (int32_t i = 0; i < weight_num; i++) { | ||
| ctx->SaveTensorForBackward( | ||
| JUST(VectorAt(outputs, i + 1))); // cublas aux. need minus 1. idx_sum:2+2w | ||
| } | ||
| for (int32_t i = 0; i < weight_num - 1; i++) { | ||
| ctx->SaveTensorForBackward(JUST(VectorAt(outputs, i + 1 + weight_num))); // hidden. | ||
| } | ||
|
|
||
| ComposedAttrMap composed_attrs(attrs, base_attrs_); | ||
| ctx->skip_final_activation = JUST(composed_attrs.GetAttr<bool>("skip_final_activation")); | ||
| ctx->dropout_rate_list = JUST(composed_attrs.GetAttr<std::vector<float>>("dropout_rate_list")); | ||
|
|
||
| return Maybe<void>::Ok(); | ||
| } | ||
|
|
||
| Maybe<void> FusedMatmulBiasAddReluDropout::Apply( | ||
| const FusedMatmulBiasAddReluDropoutCaptureState* ctx, const TensorTuple& out_grads, | ||
| TensorTuple* in_grads) const { | ||
| int32_t weight_num = ctx->weight_num; | ||
| in_grads->resize(1 + 2 * weight_num); | ||
|
|
||
| TensorTuple hiddens(weight_num - 1); | ||
| TensorTuple weights(weight_num); | ||
| TensorTuple cublas_auxs(weight_num); | ||
| TensorTuple dgrad(weight_num); | ||
|
|
||
| std::shared_ptr<one::Tensor> x = JUST(VectorAt(ctx->SavedTensors(), 0)); | ||
| std::shared_ptr<one::Tensor> out = JUST(VectorAt(ctx->SavedTensors(), 1 + weight_num)); | ||
|
|
||
| for (int32_t i = 0; i < weight_num; ++i) { | ||
| weights[i] = JUST(VectorAt(ctx->SavedTensors(), 1 + i)); | ||
| } | ||
|
|
||
| for (int32_t i = 0; i < weight_num; ++i) { | ||
| cublas_auxs[i] = JUST(VectorAt(ctx->SavedTensors(), i + 2 + weight_num)); | ||
| } | ||
|
|
||
| for (int32_t i = 0; i < weight_num - 1; ++i) { | ||
| hiddens[i] = JUST(VectorAt(ctx->SavedTensors(), i + 2 + 2 * weight_num)); | ||
| } | ||
| float rate = ctx->dropout_rate_list.at(weight_num - 1); | ||
| float scale = 0.0f; | ||
| if (rate < 1.0f) { scale = 1.0f / (1.0f - rate); } | ||
|
|
||
| /* | ||
| step1: use dy and mask to get last layer's dropout + relu grad. | ||
| Because curand_uniform distribution is (0.0, 1.0], so the value after relu will be write into mask | ||
| too. And DropoutGrad use this mask to generate grad, it will generate dropout and relu grad | ||
| simultaneously. | ||
| */ | ||
| std::shared_ptr<one::Tensor> last_bias_dy = JUST(VectorAt(out_grads, 0)); | ||
| if (!ctx->skip_final_activation || rate != 0.0f) { | ||
| last_bias_dy = JUST(functional::FusedReluDropoutGrad(JUST(VectorAt(out_grads, 0)), | ||
| cublas_auxs[weight_num - 1], scale)); | ||
| } | ||
|
|
||
| // step2: use reduce_sum to get last layer's bias grad. | ||
| std::vector<int32_t> reduce_axes_vec{0}; | ||
| if (JUST(VectorAt(ctx->biases_requires_grad, weight_num - 1))) { | ||
| JUST(VectorAt(*in_grads, 2 * weight_num)) = | ||
| JUST(functional::ReduceSum(last_bias_dy, reduce_axes_vec, false)); | ||
| } | ||
|
|
||
| std::shared_ptr<one::Tensor> cublas_dy = last_bias_dy; | ||
| for (int32_t hidden_layer_idx = weight_num - 1; hidden_layer_idx > 0; hidden_layer_idx--) { | ||
| // If it is final layer, we use out_grads[0] as dy. | ||
| if (hidden_layer_idx != weight_num - 1) { | ||
| cublas_dy = JUST(VectorAt(dgrad, hidden_layer_idx + 1)); | ||
| } | ||
| rate = ctx->dropout_rate_list.at(hidden_layer_idx - 1); | ||
| scale = 1.0; | ||
| if (rate < 1.0f) { scale = 1.0f / (1.0f - rate); } | ||
| /* | ||
| Here we use cublas to compute bias + relu + matmul grad. | ||
| Then use Matmul to compute weight grad. | ||
| */ | ||
| const auto& matmul_relu_bias_bgrad = JUST(functional::CublasBiasAddReluMatmulGrad( | ||
| cublas_dy, JUST(VectorAt(weights, hidden_layer_idx)), | ||
| JUST(VectorAt(cublas_auxs, hidden_layer_idx - 1)), /*alpha=*/scale)); | ||
|
|
||
| // dgrad | ||
| dgrad.at(hidden_layer_idx) = matmul_relu_bias_bgrad->at(0); // NOLINT | ||
|
|
||
| if (JUST(VectorAt(ctx->biases_requires_grad, (hidden_layer_idx - 1)))) { | ||
| // dbias | ||
| JUST(VectorAt(*in_grads, weight_num + hidden_layer_idx)) = | ||
| matmul_relu_bias_bgrad->at(1); // NOLINT | ||
| } | ||
| // dw | ||
| if (JUST(VectorAt(ctx->weights_requires_grad, hidden_layer_idx))) { | ||
| JUST(VectorAt(*in_grads, (1 + hidden_layer_idx))) = JUST(functional::MatMul( | ||
| cublas_dy, JUST(VectorAt(hiddens, hidden_layer_idx - 1)), true, false, 1.0)); | ||
| } | ||
| } | ||
|
|
||
| // For the first layer, we need to use 2 matmul to get grads. | ||
| std::shared_ptr<one::Tensor> last_dy; | ||
| if (weight_num != 1) { | ||
| last_dy = JUST(VectorAt(dgrad, 1)); | ||
| } else { | ||
| last_dy = last_bias_dy; | ||
| } | ||
|
|
||
| if (ctx->x_requires_grad) { | ||
| // dx: | ||
| JUST(VectorAt(*in_grads, 0)) = | ||
| JUST(functional::MatMul(last_dy, JUST(VectorAt(weights, 0)), false, false, 1.0)); | ||
| } | ||
| if (JUST(VectorAt(ctx->weights_requires_grad, 0))) { | ||
| // dw: | ||
| JUST(VectorAt(*in_grads, 1)) = | ||
| JUST(functional::MatMul(last_dy, JUST(VectorAt(ctx->SavedTensors(), 0)), true, false, 1.0)); | ||
| } | ||
|
|
||
| return Maybe<void>::Ok(); | ||
| } | ||
|
|
||
| REGISTER_OP_EXPR_GRAD_FUNCTION("fused_matmul_bias_add_relu_dropout", FusedMatmulBiasAddReluDropout); | ||
|
|
||
| } // namespace one | ||
|
|
||
| } // namespace oneflow | ||
| #endif // CUDA_VERSION >= 11060 |
This file contains bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
Oops, something went wrong.