index
:
ffwd
master
Fastforward embedding inference engine
about
summary
refs
log
tree
commit
diff
log msg
author
committer
range
path:
root
/
libffwd
/
forward.c
(
unfollow
)
Commit message (
Collapse
)
Author
Files
Lines
2026-06-27
chore: clean up comment formatting and wording
Onur Gündüz
1
-26
/
+29
2026-06-25
chore: run clang-format
Onur Gündüz
1
-41
/
+40
2026-06-21
all: split Makefile and reorg library/tools
Onur Gündüz
1
-150
/
+159
2026-06-16
all: add SentencePiece XLM-R support
Onur Gündüz
1
-6
/
+9
2026-06-16
cpu: split embed.c into model.c and forward.c
Onur Gündüz
1
-639
/
+8
2026-06-16
cpu: extract late-interaction model into late.c
Onur Gündüz
1
-387
/
+30
2026-06-16
cpu: extract alloc.{c,h} and vec.c from embed.c
Onur Gündüz
1
-122
/
+4
2026-06-16
api: dedup scalar bf16-to-f32 into dtype.h
Onur Gündüz
1
-10
/
+9
2026-06-16
chore: move sources to src/ with proper names, drop qwen_ prefix
Onur Gündüz
1
-35
/
+34
2026-06-16
all: support BERT tanh-GeLU (gelu_new / gelu_pytorch_tanh)
Onur Gündüz
1
-1
/
+4
2026-06-15
cpu: add the BERT encoder family forward and loader
Onur Gündüz
1
-16
/
+229
2026-06-15
all: add CLS first-token pooling
Onur Gündüz
1
-4
/
+10
2026-06-15
all: support GTE-Qwen2 embeddings
Onur Gündüz
1
-4
/
+29
2026-06-15
api: share config parsing across backends
Onur Gündüz
1
-146
/
+2
2026-06-15
api: accept Hugging Face model weight prefix
Onur Gündüz
1
-28
/
+38
2026-06-15
late: add CUDA late-interaction rerank path (GPU encode, host MaxSim)
Onur Gündüz
1
-0
/
+10
2026-06-15
late: batch CPU rerank candidate encoding in one packed forward
Onur Gündüz
1
-0
/
+92
2026-06-15
chore: run clang-format
Onur Gündüz
1
-516
/
+532
2026-06-14
all: add Qwen3-Embedding-4B + fix terminal token
Onur Gündüz
1
-6
/
+1
2026-06-14
all: add support for Qwen3-Embedding + causal GQA attention
Onur Gündüz
1
-26
/
+85
2026-06-14
chore: rename project
Onur Gündüz
1
-146
/
+146
2026-06-14
dist: remove distributed layer-sharding (protocol, CLI flags, slice APIs)
Onur Gündüz
1
-42
/
+0
2026-06-14
late: batch MaxSim as grouped GEMM plus row-max reduction
Onur Gündüz
1
-2
/
+60
2026-06-14
cpu: move BF16 widening scratch into the workspace
Onur Gündüz
1
-18
/
+49
2026-06-14
chore: rename sources; binaries pplx-embed, pplx-embed-server
Onur Gündüz
1
-2
/
+2
2026-06-14
gitignore: track new binary and library names
Onur Gündüz
1
-0
/
+0
2026-06-14
kernels: drop asr_ prefix naming
Onur Gündüz
1
-2
/
+2
2026-06-14
cuda: add initial CUDA embedding backend
Onur Gündüz
1
-57
/
+1
2026-06-14
chore: address static analyzer warnings
Onur Gündüz
1
-2
/
+5
2026-06-14
late: add packed maxsim scoring
Onur Gündüz
1
-3
/
+22
2026-06-14
late: add CPU token vectors and maxsim
Onur Gündüz
1
-5
/
+188
2026-06-14
all: add validation for late-interaction model compat
Onur Gündüz
1
-1
/
+25
2026-06-14
cpu: reuse tiled attention scratch buffers
Onur Gündüz
1
-3
/
+23
2026-06-14
server: batch contextual document inference
Onur Gündüz
1
-12
/
+69
2026-06-14
dist: load distributed layer shards selectively
Onur Gündüz
1
-14
/
+46
2026-06-14
dist: add distributed layer-slice foundation
Onur Gündüz
1
-21
/
+113
2026-06-14
api: add opt-in l2 normalization helper
Onur Gündüz
1
-3
/
+31
2026-06-14
server: align embedding quantization with perplexity
Onur Gündüz
1
-23
/
+5
2026-06-14
server: implement HTTP API with libae integration
Onur Gündüz
1
-0
/
+62
2026-06-14
cpu: keep batch offsets in workspace
Onur Gündüz
1
-10
/
+38
2026-06-14
cpu: tighten workspace growth
Onur Gündüz
1
-4
/
+14
2026-06-14
cpu: reduce initial workspace capacity
Onur Gündüz
1
-1
/
+2
2026-06-14
server: report workspace memory
Onur Gündüz
1
-0
/
+58
2026-06-14
kernels,cpu: fuse tiny BF16 gate up projections
Onur Gündüz
1
-3
/
+26
2026-06-14
kernels,cpu: fuse small BF16 QKV projections
Onur Gündüz
1
-3
/
+24
2026-06-14
cpu: reuse MLP scratch buffer
Onur Gündüz
1
-6
/
+3
2026-06-14
cpu: support BF16 weights
Onur Gündüz
1
-46
/
+142
2026-06-14
cpu: fuse CPU embedding hot paths
Onur Gündüz
1
-11
/
+23
2026-06-14
cpu: validate model tensor shapes
Onur Gündüz
1
-21
/
+90
2026-06-14
cpu: harden workspace allocation
Onur Gündüz
1
-41
/
+125
[next]