Georgi Gerganov b0f27361f3 sampling : avoid expensive softmax during greedy sampling (#9605) пре 1 година
..
.gitignore 2c4f566c88 tests : gitignore ggml-common.h пре 1 година
CMakeLists.txt 0226613853 threadpool : skip polling for unused threads (#9461) пре 1 година
get-model.cpp 413e7b0559 ci : add model tests + script wrapper (#4586) пре 2 година
get-model.h 413e7b0559 ci : add model tests + script wrapper (#4586) пре 2 година
run-json-schema-to-grammar.mjs 5b7b0ac8df json-schema-to-grammar improvements (+ added to server) (#5978) пре 1 година
test-arg-parser.cpp 6262d13e0b common : reimplement logging (#9418) пре 1 година
test-autorelease.cpp f486f6e1e5 ggml : add numa options (#5377) пре 1 година
test-backend-ops.cpp a5b57b08ce CUDA: enable Gemma FA for HIP/Pascal (#9581) пре 1 година
test-barrier.cpp 0226613853 threadpool : skip polling for unused threads (#9461) пре 1 година
test-c.c fbf1ddec69 Nomic Vulkan backend (#4456) пре 1 година
test-chat-template.cpp 88954f7fbd tests : fix printfs (#8068) пре 1 година
test-double-float.cpp 370b1f7e7a ggml : minor naming changes (#8433) пре 1 година
test-grad0.cpp 424c5d00a9 ggml/examples: add backend support for numerical optimization (ggml/949) пре 1 година
test-grammar-integration.cpp df270ef745 llama : refactor sampling v2 (#9294) пре 1 година
test-grammar-parser.cpp df270ef745 llama : refactor sampling v2 (#9294) пре 1 година
test-json-schema-to-grammar.cpp df270ef745 llama : refactor sampling v2 (#9294) пре 1 година
test-llama-grammar.cpp df270ef745 llama : refactor sampling v2 (#9294) пре 1 година
test-log.cpp 6262d13e0b common : reimplement logging (#9418) пре 1 година
test-lora-conversion-inference.sh 3ba780e2a8 lora : fix llama conversion script with ROPE_FREQS (#9117) пре 1 година
test-model-load-cancel.cpp f486f6e1e5 ggml : add numa options (#5377) пре 1 година
test-opt.cpp ab336a9d5e code : normalize enum names (#5697) пре 1 година
test-quantize-fns.cpp 9bc6db28d0 ggml-quants : ternary packing for TriLMs and BitNet b1.58 (#8151) пре 1 година
test-quantize-perf.cpp 370b1f7e7a ggml : minor naming changes (#8433) пре 1 година
test-rope.cpp 42c76d1358 Threadpool: take 2 (#8672) пре 1 година
test-sampling.cpp b0f27361f3 sampling : avoid expensive softmax during greedy sampling (#9605) пре 1 година
test-tokenizer-0.cpp fa79495bb4 llama : fix pre-tokenization of non-special added tokens (#8228) пре 1 година
test-tokenizer-0.py 6fbd432211 py : logging and flake8 suppression refactoring (#7081) пре 1 година
test-tokenizer-0.sh edc29433fa tests : fix test-tokenizer-0.sh пре 1 година
test-tokenizer-1-bpe.cpp 213701b51a Detokenizer fixes (#8039) пре 1 година
test-tokenizer-1-spm.cpp 213701b51a Detokenizer fixes (#8039) пре 1 година
test-tokenizer-random.py fa79495bb4 llama : fix pre-tokenization of non-special added tokens (#8228) пре 1 година