MCPcopy Create free account

hub / github.com/BICLab/SpikingBrain-7B / functions

Functions395 in github.com/BICLab/SpikingBrain-7B

⨍ Functions 395◇ Types & classes 101↳ Endpoints 1
Methodforward
(ctx, input)
hf_7B_model/activations.py:258
Methodforward
(ctx, input)
hf_7B_model/activations.py:291
Methodforward
(ctx, x, y)
hf_7B_model/activations.py:387
Methodforward
(ctx, x, y, weight, bias)
hf_7B_model/activations.py:409
Methodforward
(self, x, position_ids)
hf_7B_model/window_attention_sdpa.py:34
Methodforward
( self, hidden_states: torch.Tensor, attention_mask: Optional[torch.Tensor] = None,
hf_7B_model/window_attention_sdpa.py:150
Methodforward
Forward pass with MetaAttention. Args: query: shape = [num_tokens, num_heads, head_size] key: shape = [num_tokens, nu
vllm_hymeta/attention/meta_attention.py:703
Methodforward
( ctx, q1, q2, k2, v2, key_cache, value_cache,
vllm_hymeta/attention/fa_kernel_metatoken.py:127
Methodforward
( ctx, q1, q2, k1, k2, v1, v2, cu_seqlen_q,
vllm_hymeta/attention/fa_kernel_metatoken.py:201
Methodforward
( self, positions: torch.Tensor, query: torch.Tensor, key: torch.Tensor, )
vllm_hymeta/model_for_7B/modeling_gla_swa.py:79
Methodforward
(self, x: torch.Tensor)
vllm_hymeta/model_for_7B/modeling_gla_swa.py:125
Methodforward
( self, hidden_states: torch.Tensor, positions: torch.Tensor, attn_metadata: A
vllm_hymeta/model_for_7B/modeling_gla_swa.py:204
Methodforward
( self, hidden_states: torch.Tensor, positions: torch.Tensor, kv_caches: Optio
vllm_hymeta/model_for_7B/modeling_gla_swa.py:283
Methodforward
( self, input_ids: Optional[torch.Tensor], positions: torch.Tensor, intermedia
vllm_hymeta/model_for_7B/modeling_gla_swa.py:419
Methodforward
(self, input_ids: torch.Tensor, positions: torch.Tensor, inter
vllm_hymeta/model_for_7B/modeling_gla_swa.py:569
Methodforward
(ctx, q, k, v, g, kv_caches, slot_idx, scale=None, initial_state=None, output_final_state=False, reverse=False
vllm_hymeta/model_for_7B/my_fused_recurrent.py:88
Methodforward
(self, x: torch.Tensor)
vllm_hymeta/model_for_7B/gla_attention.py:85
Methodforward
( self, hidden_states: torch.Tensor, kv_caches: Optional[GLACacheParams] = None,
vllm_hymeta/model_for_7B/gla_attention.py:237
Methodget_attn_backend_cls
(cls, selected_backend, head_size, dtype, kv_cache_dtype, block_size, use_v1,
vllm_hymeta/platform.py:21
Methodget_builder_cls
()
vllm_hymeta/attention/meta_attention.py:74
Methodget_decoder
(self)
hf_7B_VLM/SpikingBrain_VL/modeling_spikingbrain_vl.py:1499
Methodget_decoder
(self)
W8ASpike/modeling_gla_swa.py:327
Methodget_decoder
(self)
hf_7B_model/modeling_gla_swa.py:323
Methodget_impl_cls
()
vllm_hymeta/attention/meta_attention.py:66
Methodget_input_embeddings
(self)
hf_7B_VLM/SpikingBrain_VL/modeling_spikingbrain_vl.py:1077
Methodget_input_embeddings
(self)
hf_7B_VLM/SpikingBrain_VL/modeling_spikingbrain_vl.py:1484
Methodget_input_embeddings
(self)
W8ASpike/modeling_gla_swa.py:189
Methodget_input_embeddings
(self)
W8ASpike/modeling_gla_swa.py:312
Methodget_input_embeddings
(self)
hf_7B_model/modeling_gla_swa.py:187
Methodget_input_embeddings
(self)
hf_7B_model/modeling_gla_swa.py:308
Methodget_input_embeddings
( self, input_ids: torch.Tensor, )
vllm_hymeta/model_for_7B/modeling_gla_swa.py:563
Methodget_kv_cache_shape
( num_blocks: int, block_size: int, num_kv_heads: int, head_size: int, )
vllm_hymeta/attention/meta_attention.py:82
Methodget_max_length
Returns the maximum sequence length of the cached states. Cache does not have a maximum length.
W8ASpike/cache.py:120
Methodget_max_length
Returns the maximum sequence length of the cached states. Cache does not have a maximum length.
hf_7B_model/cache.py:120
Methodget_metadata_cls
()
vllm_hymeta/attention/meta_attention.py:70
Methodget_name
()
vllm_hymeta/attention/meta_attention.py:62
Methodget_output_embeddings
(self)
hf_7B_VLM/SpikingBrain_VL/modeling_spikingbrain_vl.py:1490
Methodget_output_embeddings
(self)
W8ASpike/modeling_gla_swa.py:318
Methodget_output_embeddings
(self)
hf_7B_model/modeling_gla_swa.py:314
Methodget_seqlen_agnostic_capture_inputs
(self, batch_size: int)
vllm_hymeta/model_for_7B/modeling_gla_swa.py:558
Methodget_state_cls
()
vllm_hymeta/attention/meta_attention.py:78
Methodinit_state
(self, batch_size: int)
W8ASpike/gla_attention.py:174
Methodinit_state
(self, batch_size: int)
hf_7B_model/gla_attention.py:175
Methodis_all_cross_attn_metadata_set
All attention metadata required for enc/dec cross-attention is set. Superset of encoder attention required metadata.
vllm_hymeta/attention/meta_attention.py:214
Methodis_all_encoder_attn_metadata_set
All attention metadata required for encoder attention is set.
vllm_hymeta/attention/meta_attention.py:207
Methodis_attn_weight
(name: str)
vllm_hymeta/model_for_7B/modeling_gla_swa.py:652
Methodis_densed_mlp_weight
(name: str)
vllm_hymeta/model_for_7B/modeling_gla_swa.py:613
Methodis_layer_norm_weight
(name: str)
vllm_hymeta/model_for_7B/modeling_gla_swa.py:688
Methodlayer_fn
(prefix: str)
vllm_hymeta/model_for_7B/modeling_gla_swa.py:333
Functionlayer_norm_ref
(x, weight, bias, residual=None, eps=1e-6, prenorm=False, upcast=False)
hf_7B_model/layernorm.py:23
Methodload_attn_weight
(name: str, loaded_weight: torch.Tensor, self)
vllm_hymeta/model_for_7B/modeling_gla_swa.py:655
Methodload_basic_weight
(name: str, loaded_weight: torch.Tensor, self)
vllm_hymeta/model_for_7B/modeling_gla_swa.py:692
Methodload_mlp_weight
(name: str, loaded_weight: torch.Tensor, self)
vllm_hymeta/model_for_7B/modeling_gla_swa.py:616
Methodload_weights
(self, weights: Iterable[tuple[str, torch.Tensor]])
vllm_hymeta/model_for_7B/modeling_gla_swa.py:601
Functionlogsigmoid_bwd_kernel
( x, dx, dy, T: tl.constexpr, D: tl.constexpr, BT: tl.constexpr )
W8ASpike/activations.py:108
Functionlogsigmoid_bwd_kernel
( x, dx, dy, T: tl.constexpr, D: tl.constexpr, BT: tl.constexpr )
hf_7B_model/activations.py:108
Functionlogsigmoid_fwd_kernel
( x, y, T: tl.constexpr, D: tl.constexpr, BT: tl.constexpr )
W8ASpike/activations.py:65
Functionlogsigmoid_fwd_kernel
( x, y, T: tl.constexpr, D: tl.constexpr, BT: tl.constexpr )
hf_7B_model/activations.py:65
Methodmake_empty_intermediate_tensors
( self, batch_size: int, dtype: torch.dtype, device: torch.device )
vllm_hymeta/model_for_7B/modeling_gla_swa.py:585
Methodmodel_input_names
(self)
hf_7B_VLM/SpikingBrain_VL/processing_spikingbrain_vl.py:139
Functionmy_fused_recurrent_fwd_kernel
Kernel for linear attention decoding with KV cache. This kernel computes attention for a single token using the KV cache.
vllm_hymeta/model_for_7B/my_fused_recurrent.py:10
Functionnaive_metatoken_flash_attn
( q1, q2, k1, k2, v1, v2, softmax_scale=None, num_meta_tokens=128, sliding
vllm_hymeta/attention/fa_kernel_metatoken.py:372
Methodneuronal_charge
(self,)
W8ASpike/Int2Spike/neuron.py:31
Methodneuronal_fire
(self,)
W8ASpike/Int2Spike/neuron.py:34
Methodneuronal_reset
(self,)
W8ASpike/Int2Spike/neuron.py:37
Methodpost_process_image_text_to_text
Post-process the output of the model to decode the text. Args: generated_outputs (`torch.Tensor` or `np.ndarray`):
hf_7B_VLM/SpikingBrain_VL/processing_spikingbrain_vl.py:122
Methodprefill_metadata
Returns the prefill metadata if it is set, otherwise returns None.
vllm_hymeta/attention/meta_attention.py:223
Methodprepare
(self)
vllm_hymeta/attention/meta_attention.py:421
Methodprepare_inputs_for_generation
( self, input_ids, past_key_values=None, attention_mask=None, inputs_e
hf_7B_VLM/SpikingBrain_VL/modeling_spikingbrain_vl.py:1844
Methodprepare_inputs_for_generation
( self, input_ids: torch.LongTensor = None, past_key_values: Optional[Tuple[List[torch
W8ASpike/modeling_gla_swa.py:345
Methodprepare_inputs_for_generation
( self, input_ids: torch.LongTensor = None, past_key_values: Optional[Tuple[List[torch
hf_7B_model/modeling_gla_swa.py:341
Functionregister
()
vllm_hymeta/__init__.py:2
Functionregister_model
()
vllm_hymeta/__init__.py:6
Functionrelu_bwd
(g, x)
W8ASpike/activations.py:273
Functionrelu_bwd
(g, x)
hf_7B_model/activations.py:273
Functionrms_norm_ref
(x, weight, bias, residual=None, eps=1e-6, prenorm=False, upcast=False)
hf_7B_model/layernorm.py:39
Methodset_decoder
(self, decoder)
hf_7B_VLM/SpikingBrain_VL/modeling_spikingbrain_vl.py:1496
Methodset_decoder
(self, decoder)
W8ASpike/modeling_gla_swa.py:324
Methodset_decoder
(self, decoder)
hf_7B_model/modeling_gla_swa.py:320
Methodset_input_embeddings
(self, value)
hf_7B_VLM/SpikingBrain_VL/modeling_spikingbrain_vl.py:1080
Methodset_input_embeddings
(self, value)
hf_7B_VLM/SpikingBrain_VL/modeling_spikingbrain_vl.py:1487
Methodset_input_embeddings
(self, value)
W8ASpike/modeling_gla_swa.py:192
Methodset_input_embeddings
(self, value)
W8ASpike/modeling_gla_swa.py:315
Methodset_input_embeddings
(self, value)
hf_7B_model/modeling_gla_swa.py:190
Methodset_input_embeddings
(self, value)
hf_7B_model/modeling_gla_swa.py:311
Methodset_output_embeddings
(self, new_embeddings)
hf_7B_VLM/SpikingBrain_VL/modeling_spikingbrain_vl.py:1493
Methodset_output_embeddings
(self, new_embeddings)
W8ASpike/modeling_gla_swa.py:321
Methodset_output_embeddings
(self, new_embeddings)
hf_7B_model/modeling_gla_swa.py:317
Methodstate_size
(self, **kwargs)
W8ASpike/gla_attention.py:184
Methodstate_size
(self, **kwargs)
hf_7B_model/gla_attention.py:185
Methodswap_blocks
Swap blocks between source and destination KV caches.
vllm_hymeta/attention/meta_attention.py:95
Functiontest_weight_to_bitwise
( w: torch.Tensor = None, low: int = -8, high: int = 7, size: tuple = (1024, 1024))
W8ASpike/Int2Spike/test.py:213
Methodto_legacy_cache
(self)
W8ASpike/cache.py:124
Methodto_legacy_cache
(self)
hf_7B_model/cache.py:124
Methodwhich_layer
(name: str)
vllm_hymeta/model_for_7B/modeling_gla_swa.py:607
← previous301–395 of 395, ranked by callers