aboutsummaryrefslogtreecommitdiff
path: root/embed_server.c (unfollow)
Commit message (Collapse)AuthorFilesLines
2026-06-16chore: move sources to src/ with proper names, drop qwen_ prefixOnur Gündüz1-3725/+0
2026-06-16cpu: size default threads to cgroup cpu quota, not host core countOnur Gündüz1-1/+1
2026-06-16server: register bge-base and bge-large encodersOnur Gündüz1-0/+6
2026-06-15server: serve BERT encoders (MiniLM, bge) on /v1/embeddingsOnur Gündüz1-40/+117
2026-06-15server: guard empty-chunk memcpy and check self-pipe writesOnur Gündüz1-4/+12
2026-06-15all: support GTE-Qwen2 embeddingsOnur Gündüz1-38/+41
2026-06-15server: add Qwen3 text_type query/document instruction hintOnur Gündüz1-2/+71
2026-06-15late: add CUDA late-interaction rerank path (GPU encode, host MaxSim)Onur Gündüz1-9/+75
2026-06-15late: batch CPU rerank candidate encoding in one packed forwardOnur Gündüz1-20/+23
2026-06-15late: batch MLX rerank candidate encoding in one padded passOnur Gündüz1-30/+23
2026-06-15chore: run clang-formatOnur Gündüz1-636/+664
2026-06-15server: version flags, drop cost block, tidy batch-wait noteOnur Gündüz1-30/+35
2026-06-14server: per-family embedding encoding (Qwen3 float, pplx int8)Onur Gündüz1-57/+99
2026-06-14server: add Qwen3-Embedding-8B model slotOnur Gündüz1-0/+3
2026-06-14all: add Qwen3-Embedding-4B + fix terminal tokenOnur Gündüz1-6/+8
2026-06-14all: add support for Qwen3-Embedding + causal GQA attentionOnur Gündüz1-28/+108
2026-06-14docs: fix CLI name in commentsOnur Gündüz1-4/+1
2026-06-14cli: scope GPU options to their respective build targetsOnur Gündüz1-68/+77
2026-06-14server: remove CORS supportOnur Gündüz1-11/+1
2026-06-14cli: remove --backend, use --mlx and --cuda consistentlyOnur Gündüz1-36/+7
2026-06-14server: add late-interaction rerank APIOnur Gündüz1-42/+595
2026-06-14server: tune admission deadline by backendOnur Gündüz1-2/+5
2026-06-14server: make microbatch admission deterministicOnur Gündüz1-30/+68
2026-06-14chore: rename projectOnur Gündüz1-128/+128
2026-06-14server: resolve contextual separator from vocab, test contextual pathOnur Gündüz1-1/+8
2026-06-14build: rename make targets blas->cpu, mlx->metal, shared->libOnur Gündüz1-2/+2
2026-06-14server: default micro-batch cap to 32; add scheduler sweep driverOnur Gündüz1-3/+6
2026-06-14server: do not idle-reap clients with requests in flightOnur Gündüz1-1/+4
2026-06-14chore: rename sources; binaries pplx-embed, pplx-embed-serverOnur Gündüz1-4/+339
2026-06-14gitignore: track new binary and library namesOnur Gündüz1-0/+0
2026-06-14kernels: drop asr_ prefix namingOnur Gündüz1-2/+2
2026-06-14cli: rename flags, add --stream, split serving into pplx-embed-serverOnur Gündüz1-13/+10
2026-06-14server: fix unused-rc warning in plain BLAS buildsOnur Gündüz1-1/+2
2026-06-14cuda: add server contextual pathOnur Gündüz1-1/+74
2026-06-14chore: address static analyzer warningsOnur Gündüz1-6/+28
2026-06-14server: expose worker stage timingOnur Gündüz1-13/+95
2026-06-14tokenizer: add reusable encode workspaceOnur Gündüz1-1/+10
2026-06-14server: support http keep-aliveOnur Gündüz1-11/+98
2026-06-14mlx: add opt-in 8-bit weight quantizationOnur Gündüz1-1/+15
2026-06-14server: bound mlx batches by padded token rowsOnur Gündüz1-2/+25
2026-06-14server: reject unsafe mlx model overcommitOnur Gündüz1-0/+103
2026-06-14server: simplify context separator logicOnur Gündüz1-13/+4
2026-06-14server: batch contextual document inferenceOnur Gündüz1-179/+354
2026-06-14server: batch standard requests by token budgetOnur Gündüz1-11/+64
2026-06-14server: align embedding quantization with perplexityOnur Gündüz1-13/+11
2026-06-14server: support float embedding responsesOnur Gündüz1-18/+52
2026-06-14server: add embeddings micro-batchingOnur Gündüz1-98/+343
2026-06-14server: implement HTTP API with libae integrationOnur Gündüz1-0/+1524