1be2b8c19b
ggml : revert change to ggml_cpy, add ggml_cont_Nd instead ( #3275 )
...
ggml-ci
2023-09-20 16:12:51 +03:00
2f3a46fccf
train : make KQ_pos memory buffer permanent via dummy scale op
2023-09-20 14:14:50 +03:00
54206962c7
llama : disable MPI for now
...
ggml-ci
2023-09-20 14:07:29 +03:00
e04dc51988
ggml-cuda : add rope f16, restore performance with parallel decoding ( #3272 )
...
* ggml-cuda : add rope f16, restore performance
* offload KQ_mask with all models
* fix rope shift
---------
Co-authored-by: Georgi Gerganov <ggerganov@gmail.com>
2023-09-20 14:00:28 +03:00
db0fc2da06
simple : improve comments + free batch
2023-09-20 13:54:20 +03:00
b377bf2266
simple : add parallel decoding support
2023-09-20 13:06:34 +03:00
addae65fd4
llama : improve llama_batch API + simplify parallel example
2023-09-20 11:03:18 +03:00
a1327c71c6
parallel : rename hot-plug to continuous-batching
2023-09-20 09:24:41 +03:00
e1067efbfa
llama : fix n_kv to never become 0
2023-09-20 09:17:05 +03:00
7b7472ee26
parallel : minor
2023-09-20 00:35:10 +03:00
6028879f56
parallel : print misses on each request
2023-09-19 23:50:05 +03:00
eed3fd4234
parallel : count cache misses
2023-09-19 23:47:47 +03:00
8a9aca37c1
parallel : remove question with short answers
2023-09-19 23:34:30 +03:00
4b5f3cd6bf
parallel : process system prompt once + configurable paramters + llama API
2023-09-19 17:00:42 +03:00
82e20e9ba0
parallel : remove new line from prompt
2023-09-19 13:54:41 +03:00
d37081ae5d
llama : silence errors KV cache errors
2023-09-19 13:42:59 +03:00
16090a5dde
parallel : fix sequence termination criteria
2023-09-19 13:29:29 +03:00
806d397c1a
parallel : try smaller batches when the KV cache is fragmented
2023-09-19 13:21:36 +03:00
ddad227782
llama : fix cell_max logic + rename functions
2023-09-19 13:21:12 +03:00
36714e16d0
parallel : various improvements
2023-09-19 12:29:37 +03:00
467e307931
simple : fix token counting
2023-09-19 11:45:33 +03:00
25bd254089
make : add parallel to build + fix static functions in llama.cpp
2023-09-19 11:37:02 +03:00
7e2b9974d1
ggml-cuda : update rope implementation for parallel decoding ( #3254 )
...
* ggml-cuda : update rope implementation for parallel decoding
* better solution for p0 computation
* fix rope
* simpler rope implementation
---------
Co-authored-by: Georgi Gerganov <ggerganov@gmail.com>
2023-09-19 11:31:36 +03:00
daf4c6d360
llama : fix worst case graph build
2023-09-19 11:05:08 +03:00
fa0e677820
llama : extend batch API to select which logits to output
2023-09-19 00:24:13 +03:00
897caccdf4
fixes : speculative KV cache + llama worst-case graph
2023-09-18 22:32:28 +03:00
466b513851
parallel : disable hot-plug to avoid cache fragmentation
2023-09-18 21:34:20 +03:00
0161372b9a
parallel : example for serving multiple users in parallel
2023-09-18 20:37:28 +03:00
1f17ea631c
speculative : fix KV cache management
2023-09-18 19:01:20 +03:00
7c1bdd0e8a
llama : apply K-cache roping for Falcon and Baichuan
2023-09-18 18:26:05 +03:00
0cbf3bfef8
llama : add llama_kv_cache_shift_seq + no more context swaps
2023-09-18 18:10:43 +03:00
86c90e34f5
metal : disable concurrency optimization
2023-09-18 18:00:01 +03:00
f015b26689
llama : more robust cell_max heuristic + wip shift
2023-09-18 17:15:58 +03:00
4d76d762ef
llama : extend llama_kv_cache API
2023-09-18 15:53:03 +03:00
6952a460b9
llama : add cell_max heuristic for more efficient kv_cache
2023-09-18 15:31:24 +03:00
9f42e75489
llama : add new llama_decode() API that works with llama_batch
2023-09-18 14:23:52 +03:00
58bb5110ca
Merge branch 'master' into custom-attention-mask
2023-09-18 11:15:18 +03:00
d29e76937c
llama : unified KV cache + batch inference API
2023-09-18 11:08:15 +03:00
7ddf185537
ci : switch cudatoolkit install on windows to networked ( #3236 )
2023-09-18 02:21:47 +02:00
ee66942d7e
CUDA: fix peer access logic ( #3231 )
2023-09-17 23:35:20 +02:00
fad56936d4
metal : add rope_f16 kernel + optimize cpy kernels
2023-09-17 23:39:45 +03:00
1fb033fd85
ggml : ggml_rope now takes a vector with positions instead of n_past
2023-09-17 21:17:10 +03:00
3b4bab6a38
llama : replace ggml_diag_mask_inf with ggml_add (custom -inf mask)
2023-09-17 19:42:39 +03:00
c5df72e848
tests : verify that RoPE is "additive"
2023-09-17 17:55:12 +03:00
111163e246
CUDA: enable peer access between devices ( #2470 )
2023-09-17 16:37:53 +02:00
8b428c9bc8
llama.cpp : show model size and BPW on load ( #3223 )
2023-09-17 14:33:28 +02:00
578d8c8f5c
CUDA: fix scratch malloced on non-main device ( #3220 )
2023-09-17 14:16:22 +02:00
b541b4f0b1
Enable BUILD_SHARED_LIBS=ON on all Windows builds ( #3215 )
2023-09-16 19:35:25 +02:00
5dbc2b3213
Enable build with CUDA 11.0 (make) ( #3132 )
...
* CUDA 11.0 fixes
* Cleaner CUDA/host flags separation
Also renamed GGML_ASSUME into GGML_CUDA_ASSUME
2023-09-16 16:55:43 +02:00
b08e75baea
Fixing the last deviations from sentencepiece indicated by test-tokenizer-1 ( #3170 )
...
* Fix für #2721
* Reenable tokenizer test for LLaMa
* Add `console.cpp` dependency
* Fix dependency to `common`
* Fixing wrong fix.
* Make console usage platform specific
Work on compiler warnings.
* Adapting makefile
* Remove trailing whitespace
* Adapting the other parts of the makefile
* Fix typo.
* Fixing the last deviations from sentencepiece indicated by test-tokenizer-1
* Simplify logic
* Add missing change...
* Fix ugly compiler warning
* llama_tokenize should accept strings containing NUL now
* Adding huichen's test case
2023-09-16 13:41:33 +02:00
slaren
Georgi Gerganov
Erik Scholz
Johannes Gäßler
IsaacDynamo
Vlad
goerch