Georgi Gerganov
d291c74253
llama : handle no-vocab detokenization
2024-12-18 14:02:26 +02:00
Georgi Gerganov
824fa750d4
llama : update WavTokenizer to non-causal attn
2024-12-18 14:02:26 +02:00
Georgi Gerganov
2033fb7eef
cont [no ci]
2024-12-18 14:02:26 +02:00
Georgi Gerganov
35259e5335
cont
...
ggml-ci
2024-12-18 14:02:26 +02:00
Georgi Gerganov
980d631032
llama : refactor wavtokenizer tensors
...
ggml-ci
2024-12-18 14:02:26 +02:00
Georgi Gerganov
d1ef627c51
tts : fix tensor shapes
2024-12-18 14:02:26 +02:00
Georgi Gerganov
c096bbd8dd
tts : remove hardcoded constants
...
ggml-ci
2024-12-18 14:02:25 +02:00
Georgi Gerganov
e70f140c04
tts : outetts-voc -> wavtokenizer-dec
2024-12-18 14:02:25 +02:00
Georgi Gerganov
befdcd2492
tts : text pre-processing
2024-12-18 14:02:25 +02:00
Georgi Gerganov
3d54be4d84
tts : update default samplers
...
ggml-ci
2024-12-18 14:02:25 +02:00
Georgi Gerganov
1d7c27ca93
tts : fixes
2024-12-18 14:02:25 +02:00
Georgi Gerganov
906a0edb5a
tts : fix sampling + cut initial noise
2024-12-18 14:02:24 +02:00
Georgi Gerganov
2221e54278
tts : add matchematical constant
...
ggml-ci
2024-12-18 14:02:24 +02:00
Georgi Gerganov
d4fa34bdd4
tts : add header + minor fixes
...
ggml-ci
2024-12-18 14:02:24 +02:00
Georgi Gerganov
8329e850cc
tts : minor fix
2024-12-18 14:02:24 +02:00
Georgi Gerganov
db613915de
clip : fix new conv name
2024-12-18 14:02:24 +02:00
Georgi Gerganov
b9a011e123
tts : receive input text and generate codes
2024-12-18 14:02:24 +02:00
Georgi Gerganov
191da330fc
clean-up
2024-12-18 14:02:23 +02:00
Georgi Gerganov
e52797162e
spectrum processing
2024-12-18 14:02:23 +02:00
Georgi Gerganov
5a1c98e8d2
fft
2024-12-18 14:02:23 +02:00
Georgi Gerganov
e728cfd297
compute hann window
2024-12-18 14:02:23 +02:00
Georgi Gerganov
a1f08ad338
fix n_embd + remove llama.cpp hacks
2024-12-18 14:02:23 +02:00
Georgi Gerganov
eb1b70f42a
hann window
2024-12-18 14:02:23 +02:00
Georgi Gerganov
839035d1bb
head
2024-12-18 14:02:22 +02:00
Georgi Gerganov
fe6dd5aa61
convnext
2024-12-18 14:02:22 +02:00
Georgi Gerganov
b3ba05e5bc
layer norm
2024-12-18 14:02:22 +02:00
Georgi Gerganov
435cfd788b
pos net
2024-12-18 14:02:22 +02:00
Georgi Gerganov
3046fde420
attn
2024-12-18 14:02:22 +02:00
Georgi Gerganov
13dd8941a4
resnet
2024-12-18 14:02:22 +02:00
Georgi Gerganov
3d08d62b6c
resnet conv
2024-12-18 14:02:21 +02:00
Georgi Gerganov
5296c96ca8
group norm
2024-12-18 14:02:21 +02:00
Georgi Gerganov
6ef14091c0
first conv
2024-12-18 14:02:21 +02:00
Georgi Gerganov
aac7e04953
extract features
2024-12-18 14:02:21 +02:00
Georgi Gerganov
ff2ea75fb4
wip
2024-12-18 14:02:21 +02:00
Georgi Gerganov
f169965158
llama : add OuteTTS support (wip)
2024-12-18 14:02:20 +02:00
Georgi Gerganov
e65556f174
server : do not normalize embeddings when there is no pooling
...
ggml-ci
2024-12-18 14:02:05 +02:00
Georgi Gerganov
1b18b2d7b0
server : be explicit about the pooling type in the tests
...
ggml-ci
2024-12-18 14:01:22 +02:00
Georgi Gerganov
06e85401b0
server : output embeddings for all tokens when pooling = none
...
ggml-ci
2024-12-18 14:00:50 +02:00
Georgi Gerganov
89eaf5036a
server : add "tokens" output
...
ggml-ci
2024-12-18 13:59:47 +02:00
Georgi Gerganov
152610eda9
server : output embeddings for all tokens when pooling = none ( #10861 )
...
* server : add "tokens" output
ggml-ci
* server : output embeddings for all tokens when pooling = none
ggml-ci
* server : update readme [no ci]
* server : fix spacing [no ci]
Co-authored-by: Xuan Son Nguyen <thichthat@gmail.com>
* server : be explicit about the pooling type in the tests
ggml-ci
* server : update /embeddings and /v1/embeddings endpoints
ggml-ci
* server : do not normalize embeddings when there is no pooling
ggml-ci
* server : update readme
ggml-ci
* server : fixes
* tests : update server tests
ggml-ci
* server : update readme [no ci]
* server : remove rebase artifact
---------
Co-authored-by: Xuan Son Nguyen <thichthat@gmail.com>
2024-12-18 13:01:41 +02:00
Georgi Gerganov
0e70ba686e
server : add "tokens" output ( #10853 )
...
* server : add "tokens" output
ggml-ci
* server : update readme
ggml-ci
* server : return tokens ids only if requested
ggml-ci
* tests : improve "tokens" type check
Co-authored-by: Xuan Son Nguyen <thichthat@gmail.com>
* server : remove "tokens" from the OAI endpoint
ggml-ci
---------
Co-authored-by: Xuan Son Nguyen <thichthat@gmail.com>
2024-12-18 11:05:29 +02:00
Xuan Son Nguyen
46828872c3
server : (embeddings) using same format for "input" and "content" ( #10872 )
...
* server : (embeddings) using same format for "input" and "content"
* fix test case
* handle empty input case
* fix test
2024-12-18 10:55:09 +02:00
redbeard
6b064c92b4
docs: Fix HIP (née hipBLAS) in README ( #10880 )
...
Related to #10524 / be0e350c
2024-12-18 10:35:00 +02:00
Diego Devesa
4da69d1abd
Revert "llama : add Falcon3 support ( #10864 )" ( #10876 )
...
This reverts commit 382bc7f2e8
2024-12-18 01:36:46 +01:00
DAN™
d62b532c52
Use model->gguf_kv for loading the template instead of using the C API. ( #10868 )
...
* Bump model_template to 16384 bytes to support larger chat templates.
* Use `model->gguf_kv` for efficiency.
2024-12-17 23:24:22 +01:00
Johannes Gäßler
081b29bd2a
tests: add tests for GGUF ( #10830 )
2024-12-17 19:09:35 +01:00
Georgi Gerganov
5437d4aaf5
sync : ggml
2024-12-17 18:36:02 +02:00
Georgi Gerganov
78f766768d
cmake : fix "amd64" processor string (whisper/2638)
2024-12-17 18:35:49 +02:00
gn64
8dd19a4812
vulkan : fix soft_max.comp division by zero (whisper/2633)
...
This change prevents a division by zero error when p.KY is 0.
2024-12-17 18:35:49 +02:00
Daniel Bevenius
130d0c90bd
ggml : remove return from ggml_gallocr_allocate_node (ggml/1048)
...
This commit removes the return statement from ggml_gallocr_allocate_node
function.
The motivation behind this change is to make the code more readable and
consistent.
2024-12-17 18:35:49 +02:00
