MCPcopy Create free account

hub / github.com/Luce-Org/lucebox-hub / functions

Functions2,878 in github.com/Luce-Org/lucebox-hub

⨍ Functions 2,878◇ Types & classes 420
↓ 1 callersFunctionpick_type
(name: str, scheme: str)
server/scripts/quantize_dflash_draft.py:28
↓ 1 callersFunctionpost_chat
( base_url: str, api_key: str, model: str, messages: list[dict[str, str]], max_tokens: int
harness/benchmarks/generation_benchmark.py:147
↓ 1 callersFunctionpost_json
(path: str, payload: dict, timeout: int = 900)
server/scripts/test_prefill_cache.py:46
↓ 1 callersFunctionprefill
Run prefill kernel and return first generated token id.
optimizations/megakernel/build_corpus.py:119
↓ 1 callersMethodprefill
server/test/test_server_unit.cpp:2097
↓ 1 callersMethodprefill
server/src/qwen35/qwen35_layer_split_adapter.cpp:524
↓ 1 callersMethodprefill_chunk_tokens
server/src/common/layer_split_backend.h:28
↓ 1 callersMethodprepare_full_snap
server/src/server/prefix_cache.cpp:401
↓ 1 callersMethodprepare_inline_snap
server/src/server/prefix_cache.cpp:253
↓ 1 callersFunctionpreview
(text: str, limit: int = 180)
harness/clients/summarize_backend_pair.py:208
↓ 1 callersFunctionprewarm_drafter_once
server/src/qwen3/qwen3_drafter.cpp:84
↓ 1 callersMethodprimary_gpu
server/src/placement/placement_config.h:43
↓ 1 callersFunctionprimary_output_paths
(run_dir: Path)
harness/clients/summarize_backend_pair.py:114
↓ 1 callersMethodprint_freq_analysis
server/src/common/moe_hybrid_routing_stats.cpp:122
↓ 1 callersMethodprint_ready_banner
server/test/test_server_unit.cpp:2340
↓ 1 callersMethodprint_ready_banner
server/src/common/layer_split_backend.cpp:28
↓ 1 callersFunctionprobe_anthropic_sse
(url: str, payload: dict[str, Any])
harness/client_test_runner.py:745
↓ 1 callersFunctionprobe_openai_chat
(base_url: str, *, include_long: bool = False)
harness/client_test_runner.py:610
↓ 1 callersFunctionprobe_responses_sse
(url: str, payload: dict[str, Any])
harness/client_test_runner.py:864
↓ 1 callersMethodproject_hidden_to_tokens
server/src/laguna/laguna_dflash_target.cpp:586
↓ 1 callersMethodproject_hidden_to_tokens
server/src/gemma4/gemma4_dflash_target.cpp:143
↓ 1 callersMethodproject_hidden_to_topk
server/src/qwen35/qwen35_dflash_target.cpp:672
↓ 1 callersMethodproject_hidden_to_topk
server/src/laguna/laguna_dflash_target.cpp:638
↓ 1 callersFunctionprompt_ends_in_open_think
server/src/server/http_server.cpp:98
↓ 1 callersFunctionquantize_host_f32_to_feature_type
server/src/common/dflash_feature_ring.cpp:66
↓ 1 callersFunctionqwen35_score_and_compress
server/src/qwen3/qwen3_drafter.cpp:269
↓ 1 callersFunctionqwen35moe_prefill_chunk_limit
server/src/qwen35moe/qwen35moe_backend.cpp:67
↓ 1 callersFunctionraxAddChild
Add a new child to the node 'n' representing the character 'c' and return * its new pointer, as well as the child pointer by reference. Additionally
server/src/server/rax.c:465
↓ 1 callersFunctionraxCompressNode
server/src/server/rax.c:514
↓ 1 callersFunctionraxDefragStackFree
Free the stack allocation if the iterator used heap memory. */
server/src/server/rax.c:2364
↓ 1 callersFunctionraxDefragStackInit
Initialize the private DFS stack used by the defragmentation iterator. */
server/src/server/rax.c:2310
↓ 1 callersFunctionraxDefragStackPop
Pop the current frame from the defragmentation stack. */
server/src/server/rax.c:2359
↓ 1 callersFunctionraxFreeWithCallback
Free a whole radix tree, calling the specified callback in order to * free the auxiliary data. */
server/src/server/rax.c:1498
↓ 1 callersFunctionraxIteratorCurrentParentLink
Return the pointer-to-pointer in the tree that references the element * currently selected by the iterator. For regular key nodes this is the * pare
server/src/server/rax.c:1603
↓ 1 callersFunctionraxMaterializeInlineLeaf
Convert an inline leaf value stored in 'parentlink' into a real leaf node. */
server/src/server/rax.c:212
↓ 1 callersFunctionraxRecursiveFree
This is the core of raxFree(): performs a depth-first scan of the * tree and releases all the nodes found. */
server/src/server/rax.c:1469
↓ 1 callersFunctionraxRecursiveShow
The actual implementation of raxShow(). */
server/src/server/rax.c:2640
↓ 1 callersFunctionraxRemove
Remove the specified item. Returns 1 if the item was found and * deleted, 0 otherwise. */
server/src/server/rax.c:1269
↓ 1 callersFunctionraxRemoveChild
Low level child removal from node. The new node pointer (after the child * removal) is returned. Note that this function does not fix the pointer *
server/src/server/rax.c:1211
↓ 1 callersFunctionread_counted_i32
Read a prompt file with uint32 length prefix + N int32 token IDs.
server/src/common/daemon_loop.cpp:158
↓ 1 callersFunctionread_counted_i32
(path: Path)
server/scripts/laguna_pflash_niah.py:54
↓ 1 callersFunctionread_counted_i32_file
server/test/pflash_daemon.cpp:35
↓ 1 callersFunctionread_gguf_arch
server/src/common/moe_expert_compute_ipc.cpp:581
↓ 1 callersFunctionread_i32
(path)
optimizations/spark/spark/bench.py:48
↓ 1 callersFunctionread_i64
server/src/server/disk_prefix_cache.cpp:184
↓ 1 callersMethodread_json
(self)
harness/clients/llamacpp_compat_proxy.py:399
↓ 1 callersFunctionread_meta
server/test/test_vs_oracle.cpp:38
↓ 1 callersFunctionread_prompt_text
(args)
server/scripts/phase_split_dual_gpu.py:742
↓ 1 callersFunctionread_raw_i32
(path: Path)
server/scripts/phase_split_dual_gpu.py:60
↓ 1 callersFunctionread_rope_theta_from_config
Read rope_theta from config.json in the same directory as the .safetensors file. HuggingFace models always ship config.json alongside the weights. Ret
server/src/draft/draft_safetensors_loader.cpp:465
↓ 1 callersFunctionread_sidecar_sha
Sidecar layout (extends standard sha256sum format with a validation hint): line 1: "<64-hex> <basename>\n" (sha256sum-compatible) line 2: "# size=<
server/src/common/gguf_inspect.cpp:244
↓ 1 callersFunctionread_snapshot_tensor_header_fd
server/src/qwen35/qwen35_target_shard_ipc.cpp:43
↓ 1 callersFunctionread_snapshot_tensor_header_fd
server/src/qwen35/qwen35_target_shard_ipc_daemon.cpp:69
↓ 1 callersFunctionread_stream_until_sentinel
(r_fd: int)
server/scripts/phase_split_dual_gpu.py:69
↓ 1 callersFunctionread_tensor_bytes
(path: Path, header_size: int, info: dict)
server/scripts/convert_dflash_to_gguf.py:197
↓ 1 callersFunctionread_tensor_bytes
(path: Path, header_size: int, info: dict)
server/scripts/quantize_gemma_dflash_q8.py:70
↓ 1 callersFunctionread_tensor_bytes
(path: Path, header_size: int, info: dict)
server/scripts/quantize_draft_q8.py:99
↓ 1 callersFunctionread_u16
server/src/server/disk_prefix_cache.cpp:185
↓ 1 callersFunctionread_uncounted_i32_local
server/src/laguna/laguna_backend.cpp:1562
↓ 1 callersFunctionrecall_at_k
(net, Xte, Yte, ks=(8, 16, 24))
optimizations/spark/spark/train_pregate.py:87
↓ 1 callersMethodrecord_perf
server/src/server/server_status.h:116
↓ 1 callersFunctionregression_delta
Per-metric (baseline, now, delta, pct, regressed?) + an overall verdict.
server/scripts/profile.py:358
↓ 1 callersMethodrelease_scratch
server/src/qwen35/qwen35_backend.cpp:756
↓ 1 callersFunctionremote_hot_expert_count
server/src/common/moe_expert_compute_ipc.cpp:721
↓ 1 callersFunctionrender_tool_call_xml
server/src/server/http_server.cpp:715
↓ 1 callersFunctionreport
(model, agg)
optimizations/spark/spark/train_pregate.py:117
↓ 1 callersMethodrequest_stop
Async-signal-safe: only sets the stopping flag. The accept loop polls this flag on a short timeout, so it wakes regardless of which thread the signal
server/src/server/http_server.h:266
↓ 1 callersFunctionrequired_shared_bytes
server/src/common/target_shard_ipc.cpp:32
↓ 1 callersMethodreset_continued
Reset the continued-store tracking (call at start of each request).
server/src/server/disk_prefix_cache.h:138
↓ 1 callersMethodreset_request_state
server/test/test_server_unit.cpp:2091
↓ 1 callersMethodresident_blocks
server/src/common/kvflash_pager.h:367
↓ 1 callersFunctionresolve_chat_markers
server/src/server/prefix_cache.cpp:15
↓ 1 callersFunctionresolve_draft
(draft_dir: str, target_path: str | None = None)
server/scripts/run.py:42
↓ 1 callersFunctionresolve_draft
()
server/examples/chat.py:30
↓ 1 callersFunctionresolve_model_card
server/src/server/model_card.cpp:293
↓ 1 callersFunctionresolve_target_draft
(path: Path)
server/scripts/phase_split_dual_gpu.py:368
↓ 1 callersFunctionresolve_visible_devices
(visible_devices: str | None, fallback_device: int | None = None)
server/scripts/placement/backend_device.py:8
↓ 1 callersFunctionresponses_json
(req: dict, chat: dict, model: str)
harness/clients/llamacpp_compat_proxy.py:228
↓ 1 callersFunctionresponses_sse
(obj: dict)
harness/clients/llamacpp_compat_proxy.py:302
↓ 1 callersFunctionresponses_text
(body: dict[str, Any] | str)
harness/client_test_runner.py:562
↓ 1 callersFunctionrestore_target_cache_chain
server/src/qwen35/qwen35_target_graph.cpp:1758
↓ 1 callersFunctionrewrite_completions_to_chat
server/src/server/http_server.cpp:217
↓ 1 callersMethodrollback_to
Roll back recurrent state to position `commit_n` within the last verify batch (0-indexed). Uses SSM intermediate states captured during verify. Also t
server/src/common/dflash_target.h:69
↓ 1 callersMethodrollback_to
server/src/qwen35/qwen35_dflash_target.cpp:513
↓ 1 callersMethodrollback_to_tree
server/src/laguna/laguna_dflash_target.cpp:371
↓ 1 callersFunctionrun
(name, cfg)
server/scripts/quality_ab_simple.py:118
↓ 1 callersMethodrun_all
(self)
server/tests/test_server_comprehensive.py:1049
↓ 1 callersMethodrun_all
(self)
server/tests/test_server_smoke.py:271
↓ 1 callersMethodrun_all
(self)
server/tests/test_tokenizer.py:292
↓ 1 callersFunctionrun_ar
(path: Path, n_gen: int = N_GEN)
server/scripts/bench_llm.py:120
↓ 1 callersFunctionrun_ar
(path: Path, n_gen: int)
server/scripts/bench_agent.py:190
↓ 1 callersFunctionrun_case
( case: dict[str, Any], base_url: str, api_key: str, model: str, max_tokens: int, temp
harness/benchmarks/generation_benchmark.py:194
↓ 1 callersFunctionrun_case
server/test/test_draft_topk_cuda.cpp:46
↓ 1 callersFunctionrun_correctness
(args, tokenizer)
optimizations/megakernel/bench_pp_tg_nvfp4.py:189
↓ 1 callersFunctionrun_df
(path: Path, n_prompt, n_gen: int = N_GEN)
server/scripts/bench_llm.py:142
↓ 1 callersFunctionrun_df
(path: Path, n_prompt: int, n_gen: int, budget: int = None)
server/scripts/bench_agent.py:201
↓ 1 callersFunctionrun_gemma4_daemon
server/src/gemma4/gemma4_daemon.cpp:11
↓ 1 callersFunctionrun_gemma4_target_shard_ipc_daemon
server/src/gemma4/gemma4_layer_split_adapter.cpp:1266
↓ 1 callersFunctionrun_hot_and_shared_ffn_gpu
Fused hot routed + shared FFN in a single GPU graph compute.
server/src/common/moe_hybrid_ffn_eval.cpp:315
↓ 1 callersFunctionrun_laguna_target_shard_ipc_daemon
server/src/laguna/laguna_layer_split_adapter.cpp:1248
← previousnext →1,401–1,500 of 2,878, ranked by callers