aboutsummaryrefslogtreecommitdiff
path: root/src/mlx.c (unfollow)
Commit message (Expand)AuthorFilesLines
2026-06-21all: split Makefile and reorg library/toolsOnur Gündüz1-2783/+0
2026-06-18server: serve models by explicit labelsOnur Gündüz1-1/+2
2026-06-17all: accept F16 safetensors weights (widen to F32 at load)Onur Gündüz1-2/+20
2026-06-16all: add SentencePiece XLM-R supportOnur Gündüz1-4/+16
2026-06-16chore: move sources to src/ with proper names, drop qwen_ prefixOnur Gündüz1-3/+3
2026-06-16all: support BERT tanh-GeLU (gelu_new / gelu_pytorch_tanh)Onur Gündüz1-1/+42
2026-06-16mlx: upcast BERT norm/bias/embedding fields for BF16 modelsOnur Gündüz1-0/+13
2026-06-15mlx: add the BERT encoder family forward and loaderOnur Gündüz1-7/+269
2026-06-15all: add CLS first-token poolingOnur Gündüz1-12/+15
2026-06-15all: support GTE-Qwen2 embeddingsOnur Gündüz1-8/+40
2026-06-15api: share config parsing across backendsOnur Gündüz1-125/+2
2026-06-15api: accept Hugging Face model weight prefixOnur Gündüz1-17/+26
2026-06-15late: restore MLX device select/concat used by interaction checkOnur Gündüz1-0/+101
2026-06-15late: batch MLX rerank candidate encoding in one padded passOnur Gündüz1-74/+216
2026-06-15chore: run clang-formatOnur Gündüz1-543/+610
2026-06-14all: add Qwen3-Embedding-4B + fix terminal tokenOnur Gündüz1-7/+1
2026-06-14all: add support for Qwen3-Embedding + causal GQA attentionOnur Gündüz1-25/+126
2026-06-14chore: rename projectOnur Gündüz1-103/+103
2026-06-14dist: remove distributed layer-sharding (protocol, CLI flags, slice APIs)Onur Gündüz1-186/+0
2026-06-14chore: rename sources; binaries pplx-embed, pplx-embed-serverOnur Gündüz1-3/+3
2026-06-14gitignore: track new binary and library namesOnur Gündüz1-0/+0
2026-06-14kernels: drop asr_ prefix namingOnur Gündüz1-1/+1
2026-06-14mlx: mixed-precision BF16 inference (F32 residual, BF16 weights)Onur Gündüz1-7/+71
2026-06-14mlx: split ragged contextual batchesOnur Gündüz1-3/+137
2026-06-14mlx: tune ragged grouping for 4BOnur Gündüz1-5/+34
2026-06-14mlx: prepare contiguous linear weightsOnur Gündüz1-4/+21
2026-06-14mlx: optimize ragged embedding batchesOnur Gündüz1-2/+114
2026-06-14chore: address static analyzer warningsOnur Gündüz1-5/+13
2026-06-14mlx: use prefix pooling for f32 contextual spansOnur Gündüz1-32/+138
2026-06-14mlx: cache transposed linear weightsOnur Gündüz1-4/+46
2026-06-14mlx: keep late maxsim device residentOnur Gündüz1-25/+378
2026-06-14late: add mlx token vector supportOnur Gündüz1-8/+211
2026-06-14all: add validation for late-interaction model compatOnur Gündüz1-0/+23
2026-06-14mlx: add opt-in 8-bit weight quantizationOnur Gündüz1-21/+206
2026-06-14mlx: harden model and buffer validationOnur Gündüz1-46/+156
2026-06-14mlx: simplify equal-length embedding poolingOnur Gündüz1-24/+35
2026-06-14mlx: pool contextual bf16 spans before wideningOnur Gündüz1-11/+9
2026-06-14server: batch contextual document inferenceOnur Gündüz1-35/+119
2026-06-14mlx: pool contextual spans on deviceOnur Gündüz1-17/+95
2026-06-14dist: load distributed layer shards selectivelyOnur Gündüz1-15/+56
2026-06-14dist: add distributed layer-slice foundationOnur Gündüz1-207/+255
2026-06-14server: align embedding quantization with perplexityOnur Gündüz1-10/+4
2026-06-14server: implement HTTP API with libae integrationOnur Gündüz1-0/+156
2026-06-14kernels: validate real BF16 artifactsOnur Gündüz1-6/+25
2026-06-14api: simplify embedding APIOnur Gündüz1-1/+1
2026-06-14all: support batched embedding attentionOnur Gündüz1-59/+137
2026-03-28Refactor and enhance pplx_embed and pplx_embed_mlx for improved model supportOnur Gündüz1-270/+208
2026-03-28InitOnur Gündüz1-0/+465