vbatts/llama.cpp
1
0
Fork 0
Code Issues Pull requests Projects Releases Packages Wiki Activity Actions
3857 commits 380 branches 3056 tags 365 MiB
Commit graph

3857 commits

This branch
This branch
All branches
Author SHA1 Message Date
Georgi Gerganov
20f1789dfb vulkan : fix build (#0) 
ggml-ci
 2024-08-27 22:41:27 +03:00
Georgi Gerganov
231cff5f6f sync : ggml 2024-08-27 22:41:27 +03:00
Xie Yanbo
3246fe84d7
Fix minicpm example directory (#9111) 2024-08-27 14:33:08 +02:00
compilade
78eb487bb0
llama : fix qs.n_attention_wv for DeepSeek-V2 (#9156) 2024-08-27 13:09:23 +03:00
Xuan Son Nguyen
a77feb5d71
server : add some missing env variables (#9116) 
* server : add some missing env variables

* add LLAMA_ARG_HOST to server dockerfile

* also add LLAMA_ARG_CONT_BATCHING
 2024-08-27 11:07:01 +02:00
CausalLM
2e59d61c1b
llama : fix ChatGLM4 wrong shape (#9194) 
This should fix THUDM/glm-4-9b-chat-1m and CausalLM/miniG
 2024-08-27 09:58:22 +03:00
Carsten Kragelund Jørgensen
75e1dbbaab
llama : fix llama3.1 rope_freqs not respecting custom head_dim (#9141) 
* fix: llama3.1 rope_freqs not respecting custom head_dim

* fix: use potential head_dim for Exaone
 2024-08-27 09:53:40 +03:00
arch-btw
ad76569f8e
common : Update stb_image.h to latest version (#9161) 
* Update stb_image.h to latest version

Fixes https://github.com/ggerganov/llama.cpp/issues/7431

* Update .ecrc
 2024-08-27 08:58:50 +03:00
slaren
7d787ed96c
ggml : do not crash when quantizing q4_x_x with an imatrix (#9192) 2024-08-26 19:44:43 +02:00
Georgi Gerganov
06658ad7c3
metal : separate scale and mask from QKT in FA kernel (#9189) 
* metal : separate scale and mask from QKT in FA kernel

* metal : ne01 check no longer necessary

* metal : keep data in local memory
 2024-08-26 18:31:02 +03:00
Georgi Gerganov
fc18425b6a
ggml : add SSM Metal kernels (#8546) 
* ggml : add ggml_ssm_conv metal impl

* ggml : add ssm_scan metal impl

ggml-ci
 2024-08-26 17:55:36 +03:00
Georgi Gerganov
879275ac98
tests : fix compile warnings for unreachable code (#9185) 
ggml-ci
 2024-08-26 16:30:25 +03:00
Georgi Gerganov
7a3df798fc
ci : add VULKAN support to ggml-ci (#9055) 2024-08-26 12:19:39 +03:00
Georgi Gerganov
e5edb210cd
server : update deps (#9183) 2024-08-26 12:16:57 +03:00
slaren
0c41e03ceb
metal : gemma2 flash attention support (#9159) 2024-08-26 11:08:59 +02:00
slaren
f12ceaca0c
ggml-ci : try to improve build time (#9160) 2024-08-26 11:03:30 +02:00
Justine Tunney
436787f170
llama : fix time complexity of string replacement (#9163) 
This change fixes a bug where replacing text in a very long string could
cause llama.cpp to hang indefinitely. This is because the algorithm used
was quadratic, due to memmove() when s.replace() is called in a loop. It
seems most search results and LLM responses actually provide the O(n**2)
algorithm, which is a great tragedy. Using a builder string fixes things
 2024-08-26 09:09:53 +03:00
Herman Semenov
93bc3839f9
common: fixed not working find argument --n-gpu-layers-draft (#9175) 2024-08-26 00:54:37 +02:00
Johannes Gäßler
f91fc5639b
CUDA: fix Gemma 2 numerical issues for FA (#9166) 2024-08-25 22:11:48 +02:00
Nexesenex
16aee45179 correction 2024-08-25 14:26:29 +02:00
Nexesenex
dd3df754b2 Bad indents and trailing whitespaces 2024-08-25 03:30:43 +02:00
Nexesenex
f63860eaac Put back ffn_down tree where it was before. 2024-08-25 03:20:29 +02:00
Nexesenex
8fc46df134 Bump a bit ffn_gate and down for some GQA<2 models 2024-08-25 03:12:29 +02:00
Nexesenex
53b8eaa316 Remove deprecated rules for token embeddings 2024-08-25 03:12:29 +02:00
Nexesenex
844d11b8f3 bad indent 2024-08-25 03:12:29 +02:00
Nexesenex
5ae59714d2 Revamp Q2_K and Q3_K quants 
Q3_K_XL takes the place of Q3_K_L.
Q3_K_L becomes intermediary between Q3_K_M and XL.
 2024-08-25 03:12:29 +02:00
Nexesenex
1bde168c07 Usage of n_head to discriminate very small models 
Of which the size is more sensitive to the non repeating tensors
 2024-08-25 03:04:17 +02:00
Nexesenex
16e9c3771a various corrections on IQ2_S+ and IQ3 quants 2024-08-25 03:04:17 +02:00
Nexesenex
380b53d061 Fix IQ4_XSR 2024-08-25 03:04:17 +02:00
Nexesenex
608108597c Ravamp attn_output 2024-08-25 03:04:17 +02:00
Nexesenex
6b5cebfb2b Revamp a bit output weight 
for more granularity in low quants.
 2024-08-25 03:04:16 +02:00
Nexesenex
f796954872 Revamp FFN down and attn_k 
And complete FFN up
Shrink a bit more non GQA models
 2024-08-25 03:04:16 +02:00
Nexesenex
596a4aec86 Readd variable attn_k, attn_q, attn_o after merge 2024-08-25 03:00:13 +02:00
Nexesenex
fb2b9ea667 Merge branch 'master' into pr/8836 2024-08-25 02:59:57 +02:00
Nexesenex
3a027b878b Revamp IQ4_XSR, remove IQ3_XXXL 2024-08-25 02:54:45 +02:00
Nexesenex
e05da54eff Overhaul of FFN, if GQA and if not 2024-08-25 02:54:45 +02:00
Nexesenex
1607a02bdd Further adjustments difquant formulas 2024-08-25 02:54:45 +02:00
Nexesenex
179ad0fad4 Little rework of the difquant formulas 2024-08-25 02:54:45 +02:00
Johannes Gäßler
e11bd856d5
CPU/CUDA: Gemma 2 FlashAttention support (#8542) 
* CPU/CUDA: Gemma 2 FlashAttention support

* apply logit_softcap to scale in kernel

* disable logit softcapping tests on Metal

* remove metal check
 2024-08-24 21:34:59 +02:00
João Dinis Ferreira
8f824ffe8e
quantize : fix typo in usage help of quantize.cpp (#9145) 2024-08-24 09:22:45 +03:00
Xuan Son Nguyen
3ba780e2a8
lora : fix llama conversion script with ROPE_FREQS (#9117) 2024-08-23 12:58:53 +02:00
piDack
a07c32ea54
llama : use F32 precision in GLM4 attention and no FA (#9130) 2024-08-23 10:27:17 +03:00
Akarshan Biswas
11b84eb457
[SYCL] Add a space to supress a cmake warning (#9133) 2024-08-22 22:09:47 +08:00
luoyu-intel
1731d4238f
[SYCL] Add oneDNN primitive support (#9091) 
* add onednn

* add sycl_f16

* add dnnl stream

* add engine map

* use dnnl for intel only

* use fp16fp16fp16

* update doc
 2024-08-22 12:50:10 +08:00
compilade
a1631e53f6
llama : simplify Mamba with advanced batch splits (#8526) 
* llama : advanced batch splits

This includes equal-sequence-length batch splits which are useful
to simplify recurrent model operators.

* llama : always make recurrent state slots contiguous

* ggml : simplify mamba operators

* llama : fix integer signedness mixing

* llama : logits_all has priority over batch->logits

Otherwise, the server embeddings tests failed.
This was likely an existing problem but was only detected here
because of an additional assertion.

* llama : apply suggestions

Co-authored-by: Georgi Gerganov <ggerganov@gmail.com>

* llama : fix t5 segfault

* llama : fix Mamba session save and restore

* llama : minor cosmetic changes

* llama : rename llama_reorder_outputs to llama_output_reorder

Also move it closer to llama_output_reserve.

* llama : fix pooled embeddings when using batches with equal_seqs

* minor : add struct members for clarity

ggml-ci

* llama : fix T5 segfault again

* llama : fix Mamba pooled embeddings with multiple sequences

Until the pooled embeddings are refactored to allow splitting
across ubatches for causal embeddings,
recurrent models can only process a single sequence per ubatch
when calculating pooled embeddings.

* llama : add llama_model_is_recurrent to simplify figuring that out

This will make it easier to more cleanly support RWKV-v6 and Mamba-2.

* llama : fix simple splits when the batch contains embeddings

---------

Co-authored-by: Georgi Gerganov <ggerganov@gmail.com>
 2024-08-21 17:58:11 -04:00
Nexesenex
644aa9fd41 Correction too small tensor embeddings to quantize 
IQ2_XS doesn't seem to work as such, back to IQ2_S
 2024-08-21 13:07:32 +02:00
Nexesenex
32f6ead0d9 Improve IQ1 and IQ2 quants 
And fix mistakes for the attn.output of IQ2_XL and the ffn gate and up of IQ2_XS

Reformat attn_ouput mess and split GQA4/GQA2
 2024-08-21 12:52:45 +02:00
Nexesenex
d7b9d214fb Shrink a bit IQ3_XXS, bump a bit IQ3_M 2024-08-21 12:49:40 +02:00
Nexesenex
dbadcdd5cf harmonize formatting of tensor type conditions 2024-08-21 12:30:38 +02:00
Nexesenex
ce86019770 change function use_*_bits into difquant_*_tensors 
this to clarify what it does, especially with the 5 additional levels of difquant
 2024-08-21 12:26:12 +02:00