MCPcopy Create free account

hub / github.com/Gen-Verse/OpenClaw-RL / functions

Functions10,561 in github.com/Gen-Verse/OpenClaw-RL

Methodforward
Per-chunk layer norm. Args: input (Tensor): Input chunks, concatenated into a single tensor. Returns: Output
Megatron-LM/megatron/core/models/retro/encoder_attention.py:207
Methodforward
Cross attention for Retro decoder. Notation: ns : Sequence length. bs : Batch size. d : Hidden size.
Megatron-LM/megatron/core/models/retro/decoder_attention.py:87
Methodforward
Retro decoder bias-dropout-add. Args: training (bool): If training, then apply dropout. fused (bool): Fuse bias-dropo
Megatron-LM/megatron/core/models/retro/decoder_attention.py:305
Methodforward
RetroModel forward method. Foward input tokens & mask, along with neighbor tokens & mask, through the Retro model.. Args:
Megatron-LM/megatron/core/models/retro/model.py:23
Methodforward
Forward pass of the L2Norm module. Args: x (torch.Tensor): Input tensor. Returns: torch.Tensor: L2-
Megatron-LM/megatron/core/transformer/torch_norm.py:86
Methodforward
Replay the forward graph of the passed runner.
Megatron-LM/megatron/core/transformer/cuda_graphs.py:447
Methodforward
Perform a forward pass through the transformer layer. This method calls the core computation of a transformer layer, including
Megatron-LM/megatron/core/transformer/transformer_layer.py:464
Methodforward
Forward pass for the native attention function with context parallelism
Megatron-LM/megatron/core/transformer/dot_product_attention_context_parallel.py:154
Methodforward
(self, x, *args, **kwargs)
Megatron-LM/megatron/core/transformer/identity_op.py:13
Methodforward
(self, *args, **kwargs)
Megatron-LM/megatron/core/transformer/identity_op.py:27
Methodforward
Perform a forward pass through the attention module. Args: hidden_states (Tensor): Hidden states. attention_
Megatron-LM/megatron/core/transformer/attention.py:693
Methodforward
Perform the forward pass through the transformer block. This method handles the core computation of the transformer, including
Megatron-LM/megatron/core/transformer/transformer_block.py:588
Methodforward
Perform the forward pass through the MLP block.
Megatron-LM/megatron/core/transformer/mlp.py:140
Methodforward
Preserve the mtp by storing it in the context to avoid garbage collection. Args: output (torch.Tensor): The output tensor.
Megatron-LM/megatron/core/transformer/multi_token_prediction.py:555
Methodforward
Execute the forward pass through the Multi-Token Prediction (MTP) layer. Args: input_ids (Tensor): Input token IDs .
Megatron-LM/megatron/core/transformer/multi_token_prediction.py:913
Methodforward
Perform the forward pass through all of the MTP modules. Args: hidden_states (Tensor): Hidden states for input token wit
Megatron-LM/megatron/core/transformer/multi_token_prediction.py:1129
Methodforward
Forward pass for multi-latent attention
Megatron-LM/megatron/core/transformer/multi_latent_attention.py:202
Methodforward
Forward.
Megatron-LM/megatron/core/transformer/dot_product_attention.py:148
Methodforward
Forward of TELayerNormColumnParallelLinearGathered
Megatron-LM/megatron/core/transformer/heterogeneous/linear_replacements.py:71
Methodforward
Preserve the indexer_loss by storing it in the context to avoid garbage collection. Args: output: The output tensor (activation).
Megatron-LM/megatron/core/transformer/experimental_attention_variant/dsa.py:265
Methodforward
Forward pass for DSA Indexer. Args: x: hidden states [seqlen, batch, hidden_size]. qr: Low-rank query tensor
Megatron-LM/megatron/core/transformer/experimental_attention_variant/dsa.py:619
Methodforward
Forward pass for Sparse Attention. Args: query: Query tensor [sq, b, np, hn]. key: Key tensor [skv, b, np, h
Megatron-LM/megatron/core/transformer/experimental_attention_variant/dsa.py:732
Methodforward
Preserve the aux_loss by storing it in the context to avoid garbage collection. Args: output (torch.Tensor): The output tensor.
Megatron-LM/megatron/core/transformer/moe/moe_utils.py:180
Methodforward
Forward pass returns random logits with rank-specific seed.
Megatron-LM/megatron/core/transformer/moe/moe_utils.py:931
Methodforward
Forward pass of the RouterGatingLinearFunction function.
Megatron-LM/megatron/core/transformer/moe/moe_utils.py:960
Methodforward
Forward function
Megatron-LM/megatron/core/transformer/moe/shared_experts.py:123
Methodforward
Forward step of the GroupedMLP.
Megatron-LM/megatron/core/transformer/moe/experts.py:234
Methodforward
Forward of TEGroupedMLP Args: permuted_local_hidden_states (torch.Tensor): The permuted input hidden states of the lo
Megatron-LM/megatron/core/transformer/moe/experts.py:846
Methodforward
Forward step of the SequentialMLP.
Megatron-LM/megatron/core/transformer/moe/experts.py:1098
Methodforward
Forward method for the MoE layer.
Megatron-LM/megatron/core/transformer/moe/moe_layer.py:87
Methodforward
Forward pass for the MoE layer. The forward pass comprises four main steps: 1. Routing & Preprocessing: Route tokens to the assigned
Megatron-LM/megatron/core/transformer/moe/moe_layer.py:273
Methodforward
Forward pass of the router. Args: input (torch.Tensor): Input tensor.
Megatron-LM/megatron/core/transformer/moe/router.py:116
Methodforward
Forward pass of the router. Args: input (torch.Tensor): Input tensor.
Megatron-LM/megatron/core/transformer/moe/router.py:554
Methodforward
Forward pass of fused dispatch.
Megatron-LM/megatron/core/transformer/moe/fused_a2a.py:73
Methodforward
Forward pass of fused combine.
Megatron-LM/megatron/core/transformer/moe/fused_a2a.py:167
Methodforward
Forward pass of fused dispatch of the HybridEP backend
Megatron-LM/megatron/core/transformer/moe/fused_a2a.py:339
Methodforward
Forward pass of fused combine of the HybridEP backend
Megatron-LM/megatron/core/transformer/moe/fused_a2a.py:416
Methodforward
Perform a forward pass through the GDN module. Args: hidden_states (Tensor): Hidden states. attention_mask (
Megatron-LM/megatron/core/ssm/gated_delta_net.py:246
Methodforward
Forward function of the MambaStack class. It either returns the Loss values if labels are given or the final hidden unit
Megatron-LM/megatron/core/ssm/mamba_block.py:201
Methodforward
hidden_states: (nL, B, D) / (L B D) Returns: same shape as hidden_states
Megatron-LM/megatron/core/ssm/mamba_mixer.py:401
Methodforward
Perform a forward pass through the Mamba layer. This method implements the core computation of a Mamba layer, including the
Megatron-LM/megatron/core/ssm/mamba_layer.py:91
Methodforward
(self, x)
Megatron-LM/tests/unit_tests/test_fp8_utils.py:20
Methodforward
(self, x)
Megatron-LM/tests/unit_tests/test_layer_wise_optimizer.py:42
Methodforward
(self, x)
Megatron-LM/tests/unit_tests/test_layer_wise_optimizer.py:58
Methodforward
(self, x)
Megatron-LM/tests/unit_tests/test_optimizer_cpu_offloading.py:32
Methodforward
(self, x)
Megatron-LM/tests/unit_tests/test_muon_optimizer.py:35
Methodforward
( ctx, hidden: torch.Tensor, weight: torch.Tensor, labels: tor
Megatron-LM/tests/unit_tests/fusions/test_fused_linear_cross_entropy.py:596
Methodforward
( ctx, hidden: torch.Tensor, weight: torch.Tensor, labels: tor
Megatron-LM/tests/unit_tests/fusions/test_fused_linear_cross_entropy.py:1048
Methodforward
(self, x)
Megatron-LM/tests/unit_tests/distributed/test_mcore_fully_sharded_data_parallel.py:32
Methodforward
(self, x)
Megatron-LM/tests/unit_tests/distributed/test_mcore_fully_sharded_data_parallel.py:51
Methodforward
(self, x)
Megatron-LM/tests/unit_tests/distributed/test_distributed_data_parallel.py:23
Methodforward
Toy forward pass for the CNN, where input and output shapes match.
Megatron-LM/tests/unit_tests/distributed/fsdp/test_mfsdp_fully_shard.py:80
Methodforward
(self, x, y)
Megatron-LM/tests/unit_tests/distributed/fsdp/test_mfsdp_fully_shard.py:103
Methodforward
(self, x)
Megatron-LM/tests/unit_tests/distributed/fsdp/test_mfsdp_fully_shard.py:124
Methodforward
(self, encoder_inputs: Dict[str, Any], seq_lengths: Optional[torch.Tensor] = None)
Megatron-LM/tests/unit_tests/models/test_mimo_submodules.py:42
Methodforward
Input: (B, F, 3, 336, 336) if video, else (B, 3, 336, 336) or (num_frames, 3, 336, 336).
Megatron-LM/examples/mimo/model_providers/hf_clip_encoder.py:30
Methodforward
input_features: torch.Tensor input audio features seq_lengths: torch.Tensor the number of audio tokens corres
Megatron-LM/examples/mimo/model_providers/hf_whisper_encoder.py:13
Methodforward
Regular TEDotProductAttention + zero-out dummy attention heads.
Megatron-LM/examples/multimodal/nvlm/internvit.py:215
Methodforward
( ctx, hidden_states: torch.Tensor, w1: torch.Tensor, topk_weights: torch.Tens
slime/tests/test_fused_experts_backward.py:19
Methodforward
( ctx, intermediate_cache2: torch.Tensor, w2: torch.Tensor, topk_weights: torc
slime/tests/test_fused_experts_backward.py:133
Methodforward
( self, hidden_states: torch.Tensor, attention_mask: torch.Tensor, key_value_s
slime/slime_plugins/megatron_bridge/qwen35_vl.py:141
Methodforward
( self, position_ids: torch.Tensor, mrope_section: List[int] | None = None, pa
slime/slime_plugins/megatron_bridge/qwen35_vl.py:257
Methodforward
( self, input_ids: Tensor, position_ids: Tensor, attention_mask: Tensor,
slime/slime_plugins/megatron_bridge/qwen35_vl.py:343
Methodforward
( self, input_ids: torch.Tensor, position_ids: torch.Tensor = None, attention_
slime/slime_plugins/megatron_bridge/qwen35_vl.py:717
Methodforward
( self, hidden_states: torch.Tensor, cu_seqlens: torch.Tensor = None, )
slime/slime_plugins/models/qwen3_next.py:105
Methodforward
(ctx, x, group)
slime/slime_plugins/models/hf_attention.py:59
Methodforward
( self, hidden_states: torch.Tensor, attention_mask: torch.Tensor, key_value_s
slime/slime_plugins/models/hf_attention.py:96
Methodforward
( self, hidden_states: torch.Tensor, cu_seqlens: torch.Tensor = None, )
slime/slime_plugins/models/qwen3_5.py:84
Methodforward
(ctx, vocab_parallel_logits: torch.Tensor, process_group: dist.ProcessGroup)
slime/slime/utils/ppo_utils.py:165
Methodforward
( self, input_ids, positions, forward_batch, input
slime/slime/backends/sglang_utils/qwen3_5.py:126
Methodforward
( self, input_: torch.Tensor, weight: torch.Tensor | None = None, runtime_gath
slime/slime/backends/megatron_utils/model_provider.py:44
Methodforward
( ctx, hidden_states: torch.Tensor, w1: torch.Tensor, topk_weights: torch.Tens
slime/slime/backends/fsdp_utils/kernels/fused_experts.py:17
Methodforward
(ctx, intermediate_cache1: torch.Tensor)
slime/slime/backends/fsdp_utils/kernels/fused_experts.py:175
Methodforward
( ctx, intermediate_cache2: torch.Tensor, w2: torch.Tensor, topk_weights: torc
slime/slime/backends/fsdp_utils/kernels/fused_experts.py:205
Methodforward
( ctx, intermediate_cache3: torch.Tensor, hidden_states_shape, )
slime/slime/backends/fsdp_utils/kernels/fused_experts.py:362
Methodforward
(self, hidden_states: torch.Tensor)
slime/slime/backends/fsdp_utils/models/qwen3_moe.py:90
Functionforward_backward_no_pipelining
Run forward and backward passes with no pipeline parallelism
Megatron-LM/megatron/core/pipeline_parallel/schedules.py:512
Functionforward_backward_pipelining_with_interleaving
Run interleaved 1F1B schedule (model split into model chunks), with communication between pipeline stages as needed. Returns dictionary with
Megatron-LM/megatron/core/pipeline_parallel/schedules.py:846
Functionforward_backward_pipelining_without_interleaving
Run non-interleaved 1F1B schedule, with communication between pipeline stages. Returns dictionary with losses if the last stage, empty dict otherw
Megatron-LM/megatron/core/pipeline_parallel/schedules.py:1989
Methodforward_by_lease
( self, global_lease: str, path: str, payload: dict[str, Any], timeout
terminal-rl/router_server.py:154
Functionforward_dp_epilogue
forward epilogue in dp
Megatron-LM/megatron/core/fusions/linear_cross_entropy/blackwell/triton.py:48
Methodforward_feature_maps
(self, x)
Megatron-LM/megatron/legacy/model/vision/esvit_swin_backbone.py:614
Methodforward_hook
(_module, inputs, output)
Megatron-LM/megatron/core/distributed/fsdp/src/megatron_fsdp/megatron_fsdp.py:821
Methodforward_impl
forward pass for pre-processing. This method handles: 1. Decoder embedding computation 2. Rotary positional embedding computa
Megatron-LM/megatron/core/models/gpt/fine_grained_callables.py:108
Methodforward_impl
Implements the forward pass for postprocessing. This method handles: 1. Output layer computation 2. Loss computation if label
Megatron-LM/megatron/core/models/gpt/fine_grained_callables.py:161
Methodforward_impl
Calls the submodule as the forward pass.
Megatron-LM/megatron/core/models/gpt/fine_grained_callables.py:273
Methodforward_model
Forward pass data, and gather parallel output tensors.
Megatron-LM/tests/functional_tests/test_cases/common/ckpt_converter/__main__.py:716
Methodforward_pass_without_pipeline_parallel
Utility to carry out simple forward pass for TP or no model parallel models Runs a very simple forward pass for model. Used in the case of m
Megatron-LM/megatron/core/inference/model_inference_wrappers/t5/t5_inference_wrapper.py:197
Methodforward_post_hook
(module, *_)
Megatron-LM/megatron/core/extensions/transformer_engine.py:1903
Methodforward_pre_hook
(module, *_)
Megatron-LM/megatron/core/extensions/transformer_engine.py:1882
Methodforward_return_n_last_blocks
(self, x, n=1, return_patch_avgpool=False, depth=[])
Megatron-LM/megatron/legacy/model/vision/esvit_swin_backbone.py:663
Methodforward_selfattention
(self, x, n=1)
Megatron-LM/megatron/legacy/model/vision/esvit_swin_backbone.py:630
Functionforward_step
Forward training step. Args: data_iterator : Input data iterator model (T5Model): The T5 Model
Megatron-LM/pretrain_t5.py:174
Functionforward_step
Forward training step. Args: data_iterator : Input data iterator model (GPTModel): The GPT Model return_schedule_plan (bo
Megatron-LM/pretrain_gpt.py:144
Functionforward_step
Forward training step. Args: data_iterator: Iterable dataset. model (megatron.core.models.multimodal.llava_model.LLaVAModel): Mul
Megatron-LM/pretrain_vlm.py:355
Functionforward_step
Forward step.
Megatron-LM/pretrain_ict.py:118
Functionforward_step
Forward training step. Args: data_iterator : Input data iterator model (MambaModel): The GPT Model
Megatron-LM/pretrain_mamba.py:122
Functionforward_step
Forward training step. Args: data_iterator : Input data iterator model (GPTModel): The GPT Model
Megatron-LM/train_rl.py:173
Functionforward_step
Forward step.
Megatron-LM/pretrain_retro.py:143
Functionforward_step
Forward step.
Megatron-LM/pretrain_vision_dino.py:65
← previousnext →7,501–7,600 of 10,561, ranked by callers