SYMBOL INDEX (139 symbols across 20 files)

FILE: bench.py
  function main (line 8) | def main():

FILE: example.py
  function main (line 6) | def main():

FILE: nanovllm/config.py
  class Config (line 7) | class Config:
    method __post_init__ (line 20) | def __post_init__(self):

FILE: nanovllm/engine/block_manager.py
  class Block (line 8) | class Block:
    method __init__ (line 10) | def __init__(self, block_id):
    method update (line 16) | def update(self, hash: int, token_ids: list[int]):
    method reset (line 20) | def reset(self):
  class BlockManager (line 26) | class BlockManager:
    method __init__ (line 28) | def __init__(self, num_blocks: int, block_size: int):
    method compute_hash (line 36) | def compute_hash(cls, token_ids: list[int], prefix: int = -1):
    method _allocate_block (line 43) | def _allocate_block(self, block_id: int) -> Block:
    method _deallocate_block (line 51) | def _deallocate_block(self, block_id: int) -> Block:
    method can_allocate (line 56) | def can_allocate(self, seq: Sequence) -> bool:
    method allocate (line 59) | def allocate(self, seq: Sequence):
    method deallocate (line 84) | def deallocate(self, seq: Sequence):
    method can_append (line 93) | def can_append(self, seq: Sequence) -> bool:
    method may_append (line 96) | def may_append(self, seq: Sequence):

FILE: nanovllm/engine/llm_engine.py
  class LLMEngine (line 15) | class LLMEngine:
    method __init__ (line 17) | def __init__(self, model, **kwargs):
    method exit (line 36) | def exit(self):
    method add_request (line 42) | def add_request(self, prompt: str | list[int], sampling_params: Sampli...
    method step (line 48) | def step(self):
    method is_finished (line 56) | def is_finished(self):
    method generate (line 59) | def generate(

FILE: nanovllm/engine/model_runner.py
  class ModelRunner (line 15) | class ModelRunner:
    method __init__ (line 17) | def __init__(self, config: Config, rank: int, event: Event | list[Even...
    method exit (line 50) | def exit(self):
    method loop (line 61) | def loop(self):
    method read_shm (line 68) | def read_shm(self):
    method write_shm (line 76) | def write_shm(self, method_name, *args):
    method call (line 85) | def call(self, method_name, *args):
    method warmup_model (line 91) | def warmup_model(self):
    method allocate_kv_cache (line 100) | def allocate_kv_cache(self):
    method prepare_block_tables (line 120) | def prepare_block_tables(self, seqs: list[Sequence]):
    method prepare_prefill (line 126) | def prepare_prefill(self, seqs: list[Sequence]):
    method prepare_decode (line 164) | def prepare_decode(self, seqs: list[Sequence]):
    method prepare_sample (line 182) | def prepare_sample(self, seqs: list[Sequence]):
    method run_model (line 190) | def run_model(self, input_ids: torch.Tensor, positions: torch.Tensor, ...
    method run (line 208) | def run(self, seqs: list[Sequence], is_prefill: bool) -> list[int]:
    method capture_cudagraph (line 217) | def capture_cudagraph(self):

FILE: nanovllm/engine/scheduler.py
  class Scheduler (line 8) | class Scheduler:
    method __init__ (line 10) | def __init__(self, config: Config):
    method is_finished (line 18) | def is_finished(self):
    method add (line 21) | def add(self, seq: Sequence):
    method schedule (line 24) | def schedule(self) -> tuple[list[Sequence], bool]:
    method preempt (line 60) | def preempt(self, seq: Sequence):
    method postprocess (line 65) | def postprocess(self, seqs: list[Sequence], token_ids: list[int]) -> l...

FILE: nanovllm/engine/sequence.py
  class SequenceStatus (line 8) | class SequenceStatus(Enum):
  class Sequence (line 14) | class Sequence:
    method __init__ (line 18) | def __init__(self, token_ids: list[int], sampling_params = SamplingPar...
    method __len__ (line 31) | def __len__(self):
    method __getitem__ (line 34) | def __getitem__(self, key):
    method is_finished (line 38) | def is_finished(self):
    method num_completion_tokens (line 42) | def num_completion_tokens(self):
    method prompt_token_ids (line 46) | def prompt_token_ids(self):
    method completion_token_ids (line 50) | def completion_token_ids(self):
    method num_cached_blocks (line 54) | def num_cached_blocks(self):
    method num_blocks (line 58) | def num_blocks(self):
    method last_block_num_tokens (line 62) | def last_block_num_tokens(self):
    method block (line 65) | def block(self, i):
    method append_token (line 69) | def append_token(self, token_id: int):
    method __getstate__ (line 74) | def __getstate__(self):
    method __setstate__ (line 78) | def __setstate__(self, state):

FILE: nanovllm/layers/activation.py
  class SiluAndMul (line 6) | class SiluAndMul(nn.Module):
    method __init__ (line 8) | def __init__(self):
    method forward (line 12) | def forward(self, x: torch.Tensor) -> torch.Tensor:

FILE: nanovllm/layers/attention.py
  function store_kvcache_kernel (line 11) | def store_kvcache_kernel(
  function store_kvcache (line 33) | def store_kvcache(key: torch.Tensor, value: torch.Tensor, k_cache: torch...
  class Attention (line 43) | class Attention(nn.Module):
    method __init__ (line 45) | def __init__(
    method forward (line 59) | def forward(self, q: torch.Tensor, k: torch.Tensor, v: torch.Tensor):

FILE: nanovllm/layers/embed_head.py
  class VocabParallelEmbedding (line 9) | class VocabParallelEmbedding(nn.Module):
    method __init__ (line 11) | def __init__(
    method weight_loader (line 27) | def weight_loader(self, param: nn.Parameter, loaded_weight: torch.Tens...
    method forward (line 34) | def forward(self, x: torch.Tensor):
  class ParallelLMHead (line 45) | class ParallelLMHead(VocabParallelEmbedding):
    method __init__ (line 47) | def __init__(
    method forward (line 56) | def forward(self, x: torch.Tensor):

FILE: nanovllm/layers/layernorm.py
  class RMSNorm (line 5) | class RMSNorm(nn.Module):
    method __init__ (line 7) | def __init__(
    method rms_forward (line 17) | def rms_forward(
    method add_rms_forward (line 29) | def add_rms_forward(
    method forward (line 42) | def forward(

FILE: nanovllm/layers/linear.py
  function divide (line 7) | def divide(numerator, denominator):
  class LinearBase (line 12) | class LinearBase(nn.Module):
    method __init__ (line 14) | def __init__(
    method forward (line 33) | def forward(self, x: torch.Tensor) -> torch.Tensor:
  class ReplicatedLinear (line 37) | class ReplicatedLinear(LinearBase):
    method __init__ (line 39) | def __init__(
    method weight_loader (line 47) | def weight_loader(self, param: nn.Parameter, loaded_weight: torch.Tens...
    method forward (line 50) | def forward(self, x: torch.Tensor) -> torch.Tensor:
  class ColumnParallelLinear (line 54) | class ColumnParallelLinear(LinearBase):
    method __init__ (line 56) | def __init__(
    method weight_loader (line 65) | def weight_loader(self, param: nn.Parameter, loaded_weight: torch.Tens...
    method forward (line 72) | def forward(self, x: torch.Tensor) -> torch.Tensor:
  class MergedColumnParallelLinear (line 76) | class MergedColumnParallelLinear(ColumnParallelLinear):
    method __init__ (line 78) | def __init__(
    method weight_loader (line 87) | def weight_loader(self, param: nn.Parameter, loaded_weight: torch.Tens...
  class QKVParallelLinear (line 96) | class QKVParallelLinear(ColumnParallelLinear):
    method __init__ (line 98) | def __init__(
    method weight_loader (line 114) | def weight_loader(self, param: nn.Parameter, loaded_weight: torch.Tens...
  class RowParallelLinear (line 131) | class RowParallelLinear(LinearBase):
    method __init__ (line 133) | def __init__(
    method weight_loader (line 142) | def weight_loader(self, param: nn.Parameter, loaded_weight: torch.Tens...
    method forward (line 149) | def forward(self, x: torch.Tensor) -> torch.Tensor:

FILE: nanovllm/layers/rotary_embedding.py
  function apply_rotary_emb (line 6) | def apply_rotary_emb(
  class RotaryEmbedding (line 17) | class RotaryEmbedding(nn.Module):
    method __init__ (line 19) | def __init__(
    method forward (line 38) | def forward(
  function get_rope (line 52) | def get_rope(

FILE: nanovllm/layers/sampler.py
  class Sampler (line 5) | class Sampler(nn.Module):
    method __init__ (line 7) | def __init__(self):
    method forward (line 11) | def forward(self, logits: torch.Tensor, temperatures: torch.Tensor):

FILE: nanovllm/llm.py
  class LLM (line 4) | class LLM(LLMEngine):

FILE: nanovllm/models/qwen3.py
  class Qwen3Attention (line 14) | class Qwen3Attention(nn.Module):
    method __init__ (line 16) | def __init__(
    method forward (line 71) | def forward(
  class Qwen3MLP (line 90) | class Qwen3MLP(nn.Module):
    method __init__ (line 92) | def __init__(
    method forward (line 112) | def forward(self, x):
  class Qwen3DecoderLayer (line 119) | class Qwen3DecoderLayer(nn.Module):
    method __init__ (line 121) | def __init__(
    method forward (line 145) | def forward(
  class Qwen3Model (line 161) | class Qwen3Model(nn.Module):
    method __init__ (line 163) | def __init__(
    method forward (line 172) | def forward(
  class Qwen3ForCausalLM (line 185) | class Qwen3ForCausalLM(nn.Module):
    method __init__ (line 194) | def __init__(
    method forward (line 204) | def forward(
    method compute_logits (line 211) | def compute_logits(

FILE: nanovllm/sampling_params.py
  class SamplingParams (line 5) | class SamplingParams:
    method __post_init__ (line 10) | def __post_init__(self):

FILE: nanovllm/utils/context.py
  class Context (line 6) | class Context:
  function get_context (line 18) | def get_context():
  function set_context (line 21) | def set_context(is_prefill, cu_seqlens_q=None, cu_seqlens_k=None, max_se...
  function reset_context (line 25) | def reset_context():

FILE: nanovllm/utils/loader.py
  function default_weight_loader (line 8) | def default_weight_loader(param: nn.Parameter, loaded_weight: torch.Tens...
  function load_model (line 12) | def load_model(model: nn.Module, path: str):