História revízii

Autor SHA1 Správa Dátum
  compilade 557410b8f0 llama : greatly reduce output buffer memory usage (#6122) 1 rok pred
  slaren f30ea47a87 llama : add pipeline parallelism support (#6017) 1 rok pred
  Georgi Gerganov 05b06210c9 llama : more consistent names of count variables (#5994) 1 rok pred
  slaren d894f352bf perplexity : support using multiple sequences to allow larger batch sizes (#5946) 1 rok pred
  compilade c2101a2e90 llama : support Mamba Selective State Space Models (#5328) 1 rok pred
  Georgi Gerganov b1de96824b ci : fix wikitext url + compile warnings (#5569) 1 rok pred
  Herman Semenov 5d3de51f97 ggml, common, examples, tests : fixed type arguments in printf (#5528) 1 rok pred
  bmwl f486f6e1e5 ggml : add numa options (#5377) 1 rok pred
  Michael Klimenko 52bb63c708 refactor : switch to emplace_back to avoid extra object (#5291) 1 rok pred
  kalomaze 191221178f perplexity : fix KL divergence calculations on Windows (#5273) 1 rok pred
  Kawrakow 44879ee885 Additional KL-divergence statistics (#5081) 2 rokov pred
  Georgi Gerganov 89758723c7 minor : clean-up some warnings and style (#5094) 2 rokov pred
  Kawrakow 6f9939d119 KL-divergence (#5076) 2 rokov pred
  Kawrakow 7dcbe39d36 Add ability to evauate multiple choice tasks (#5047) 2 rokov pred
  Jared Van Bortel 97c1549808 perplexity : fix MSVC build after #5020 (#5043) 2 rokov pred
  Kawrakow 7051aacfac winogrande: evaluate log-probs in parallel (#5036) 2 rokov pred
  Kawrakow 993fba8180 perplexity: avoid unnecessary alloocations and logit copies (#5035) 2 rokov pred
  Georgi Gerganov 8b20858e5e perplexity : faster Winogrande via batching (#5024) 2 rokov pred
  Georgi Gerganov d391ae9b49 perplexity : fix winogrande N tasks option 2 rokov pred
  Kawrakow 3e945cc1e9 HellaSwag: speed up by parallelizing log-prob evaluation (#5020) 2 rokov pred
  Georgi Gerganov ad19812cda perplexity : faster HellaSwag via batching (#5017) 2 rokov pred
  Kawrakow 682986a08e Add Winogrande evaluation (#5015) 2 rokov pred
  Georgi Gerganov 959ef0c0df perplexity : fix kv cache handling for hellaswag (#4981) 2 rokov pred
  Kerfuffle 91f6499393 Respect tokenizer.ggml.add_bos_token value when tokenizing (#4040) 2 rokov pred
  cebtenzzre b12fa0d1c1 build : link against build info instead of compiling against it (#3879) 2 rokov pred
  Kerfuffle 6e08281e58 Extend llama_kv_cache_seq_rm to allow matching any sequence (#3843) 2 rokov pred
  Marcus Dunn 5be6c803fa llama : remove token functions with `context` args in favor of `model` (#3720) 2 rokov pred
  slaren 16bc66d947 llama.cpp : split llama_context_params into model and context params (#3301) 2 rokov pred
  Georgi Gerganov ec893798b7 llama : custom attention mask + parallel decoding + no context swaps (#3228) 2 rokov pred
  Cebtenzzre 8781013ef6 make : restore build-info.h dependency for several targets (#3205) 2 rokov pred