Skip to content

Actions: max-krasnyansky/llama.cpp

Actions

Check Pre-Tokenizer Hashes

Actions

Loading...
Loading

Show workflow options

Create status badge

Loading
pre-tokenizer-hashes.yml
21 workflow runs
21 workflow runs

Filter by Event

⌥ + click/return to exclude

Filter by Status

⌥ + click/return to exclude

Filter by Branch

⌥ + click/return to exclude

Filter by Actor

⌥ + click/return to exclude
CUDA: also store node->src ne/nb for graph equality (#21736) Check Pre-Tokenizer Hashes #21: Commit a29e4c0 pushed by max-krasnyansky
1m 0s master
master
1m 0s
models : move the token embedding norms to the first layer (#20943) Check Pre-Tokenizer Hashes #20: Commit 9f102a1 pushed by max-krasnyansky
3m 20s master
master
3m 20s
hex-rope: use indices and unroll the loops Check Pre-Tokenizer Hashes #19: Commit f9d5f9d pushed by max-krasnyansky
1m 4s master
master
1m 4s
hexagon: refactor unary ops to use local context struct and DMA/VTCM Check Pre-Tokenizer Hashes #18: Commit 0d99b35 pushed by max-krasnyansky
5m 25s master
master
5m 25s
hexagon: refactor set/get/sum-rows ops to use local context Check Pre-Tokenizer Hashes #17: Commit af7c46c pushed by max-krasnyansky
2h 10m 24s master
master
2h 10m 24s
hexagon: fix editor config failures Check Pre-Tokenizer Hashes #16: Commit 51a9b96 pushed by max-krasnyansky
1m 9s master
master
1m 9s
hexagon: fix editor config check Check Pre-Tokenizer Hashes #15: Commit 5ff7954 pushed by max-krasnyansky
1m 6s master
master
1m 6s
ggml-cpu: use LUT for converting e8->f32 scales on x86 (#19288) Check Pre-Tokenizer Hashes #14: Commit 2ceda3f pushed by max-krasnyansky
59s master
master
59s
hexagon: remove _opt variants of add/sub/mul as they simply fully ali… Check Pre-Tokenizer Hashes #13: Commit dbeb27e pushed by max-krasnyansky
30m 43s hexagon-host-buffs
hexagon-host-buffs
30m 43s
hexagon: remove _opt variants of add/sub/mul as they simply fully ali… Check Pre-Tokenizer Hashes #12: Commit dbeb27e pushed by max-krasnyansky
31s master
master
31s
hexagon: workgin CPY fp16 -> fp32 Check Pre-Tokenizer Hashes #11: Commit 4bf921f pushed by max-krasnyansky
34m 18s hexagon-host-buffs
hexagon-host-buffs
34m 18s
hexagon: workgin CPY fp16 -> fp32 Check Pre-Tokenizer Hashes #10: Commit 4bf921f pushed by max-krasnyansky
9m 40s master
master
9m 40s
hexagon: add support for OP_CPY Check Pre-Tokenizer Hashes #9: Commit 5064ac7 pushed by max-krasnyansky
34s master
master
34s
hexagon: update adb/run-bench.sh to properly support experimental and… Check Pre-Tokenizer Hashes #8: Commit e12c475 pushed by max-krasnyansky
31s master
master
31s
hexagon: add support for set-rows fp32 -> fp16 with i32/i64 row-idx Check Pre-Tokenizer Hashes #7: Commit a638805 pushed by max-krasnyansky
28s master
master
28s
hexagon: improve fp16 matmul and add fp32/fp16 flash-attention Check Pre-Tokenizer Hashes #6: Commit 52a8ec4 pushed by max-krasnyansky
33s master
master
33s
hexagon: improv fp16 matmul and add basic fp32/fp16 flash-attention Check Pre-Tokenizer Hashes #5: Commit c63b663 pushed by max-krasnyansky
32s master
master
32s
hex-fa: optimize and fix hvx_mad_f32 Check Pre-Tokenizer Hashes #4: Commit f61e676 pushed by max-krasnyansky
29s hexagon-flash-atten
hexagon-flash-atten
29s
Webui/prompt processing progress (#18300) Check Pre-Tokenizer Hashes #3: Commit c9a3b40 pushed by max-krasnyansky
29s master
master
29s
server: return_progress to also report 0% processing state (#18305) Check Pre-Tokenizer Hashes #2: Commit 5ee4e43 pushed by max-krasnyansky
27m 35s master
master
27m 35s
gelu: fix N+2 block prefetch Check Pre-Tokenizer Hashes #1: Commit 4ad1507 pushed by max-krasnyansky
28s hexagon-gelu-optimization
hexagon-gelu-optimization
28s