Commit graph

668 commits

Author SHA1 Message Date
Andrei Betlen
47bad30dd7 fix: LlamaHFTokenizer now receives pre_tokens 2024-02-23 12:23:24 -05:00
Andrei Betlen
ded5d627a5 chore: Bump version 2024-02-23 11:32:43 -05:00
Luke Stanley
858496224e
feat: Auto detect Mixtral's slightly different format (#1214) 2024-02-23 11:27:38 -05:00
Andrei Betlen
db776a885c fix: module 'llama_cpp.llama_cpp' has no attribute 'c_uint8' 2024-02-23 11:24:53 -05:00
Andrei Betlen
427d816ebf chore: Bump version 2024-02-23 04:54:08 -05:00
Alvaro Bartolome
251a8a2cad
feat: Add Google's Gemma formatting via chat_format="gemma" (#1210)
* Add Google's Gemma formatting via `chat_format="gemma"`

* Replace `raise ValueError` with `logger.debug`

Co-authored-by: Andrei <abetlen@gmail.com>

---------

Co-authored-by: Andrei <abetlen@gmail.com>
2024-02-23 04:40:52 -05:00
Andrei Betlen
b9aca612af misc: use typesafe byref for internal classes 2024-02-23 03:40:07 -05:00
Andrei Betlen
a0ce429dc0 misc: use decorator to bind low level api functions, fixes docs 2024-02-23 03:39:38 -05:00
Andrei Betlen
e10af30cf1 fix: TypeAlias import error 2024-02-22 03:27:28 -05:00
Andrei Betlen
3561ebf536 Merge branch 'main' of https://github.com/abetlen/llama-cpp-python into main 2024-02-22 03:25:13 -05:00
Andrei Betlen
aefcb8f71a misc: additional type annotations for low level api 2024-02-22 02:00:09 -05:00
Andrei Betlen
3921e10770 feat: support minItems/maxItems in JSON grammar converter (by @nopperl) 2024-02-22 00:17:06 -05:00
Andrei Betlen
e6d6260a91 fix: Update from_pretrained defaults to match hf_hub_download 2024-02-22 00:10:23 -05:00
Andrei Betlen
dd22010e85 fix: Raise exceptions when llama model or context fails to load 2024-02-22 00:09:45 -05:00
Andrei Betlen
3632241e98 chore: Bump version 2024-02-21 23:09:13 -05:00
Andrei Betlen
0653e15c20 feat: Update llama.cpp 2024-02-21 23:04:52 -05:00
Andrei Betlen
7981e9ce1e chore: Bump version 2024-02-21 16:30:59 -05:00
Andrei
7f51b6071f
feat(low-level-api): Improve API static type-safety and performance (#1205) 2024-02-21 16:25:38 -05:00
Andrei
0f8aa4ab5c
feat: Pull models directly from huggingface (#1206)
* Add from_pretrained method to Llama class

* Update docs

* Merge filename and pattern
2024-02-21 16:25:10 -05:00
Andrei Betlen
e42f62c247 chore: Bump version 2024-02-21 11:09:40 -05:00
Andrei Betlen
4edde21b3d feat: Update llama.cpp 2024-02-21 11:05:58 -05:00
Andrei Betlen
6225f027e5 feat: Update llama.cpp 2024-02-19 04:11:34 -05:00
Andrei Betlen
748c0ce057 feat: Update llama.cpp 2024-02-18 21:30:36 -05:00
Andrei Betlen
53f6f5f415 fix: self.numa missing 2024-02-17 01:02:33 -05:00
Andrei Betlen
fdce078cb9 feat: Update llama.cpp 2024-02-17 00:37:51 -05:00
Andrei Betlen
f736827b9b chore: Bump version 2024-02-15 23:10:50 -05:00
Andrei Betlen
0ce66bc080 fix: create_embedding broken response for input type str 2024-02-15 16:09:48 -05:00
khimaros
ea1f88dd29
fix: Use '\n' seperator for EventSourceResponse (#1188)
this fixes compatibility with some OpenAI clients, including BetterChatGPT (https://github.com/ztjhz/BetterChatGPT/issues/537).

Co-authored-by: Andrei <abetlen@gmail.com>
2024-02-15 15:20:13 -05:00
Andrei Betlen
a5cfeb7763 feat: Update llama.cpp 2024-02-15 15:17:30 -05:00
Douglas Hanley
7bb91f025f
fix: Incorporate embedding pooling layer fixes (#1194)
* remove division by token count

* truncate to n_batch, not n_ctx
2024-02-15 15:16:30 -05:00
Andrei Betlen
ae71ad1a14 Bump version 2024-02-14 04:31:42 -05:00
Douglas Hanley
d7a67917ba
feat: Support batch embeddings (#1186)
* handle batched embeddings

* fix normalization issue

* fix type hints, ensure no breaking changes to embed

* Clear kv cache / reset internal state after embedding complete

---------

Co-authored-by: Andrei <abetlen@gmail.com>
2024-02-14 04:26:09 -05:00
Andrei Betlen
7b9960d1cb Update llama.cpp 2024-02-14 03:47:21 -05:00
Andrei Betlen
6943bab6d8 fix: destructor exception where internal classes are missing some uninitialized attributes 2024-02-14 03:38:41 -05:00
Andrei Betlen
07a783779a fix: Update openbuddy prompt format. Closes #1155 2024-02-13 23:57:10 -05:00
Andrei Betlen
345215a76c fix: more chatml-function-calling fixes 2024-02-13 23:02:50 -05:00
Andrei Betlen
b1637c2319 Bump version 2024-02-13 12:35:04 -05:00
Andrew Lapp
d6be5333e1
fix: sample idx off-by-one error for logit_processors (#1179)
* fix sample_idx off-by-one error

* self._scores is indexed differently, only modify the index within self._input_ids

---------

Co-authored-by: Andrew Lapp <andrew@rew.la>
Co-authored-by: Andrei <abetlen@gmail.com>
2024-02-13 12:26:07 -05:00
Andrei Betlen
f7cdf78788 Update llama.cpp 2024-02-13 12:24:00 -05:00
Andrei Betlen
68fb71b6a2 fix: missing generation_prompt in chatml-function-calling 2024-02-13 03:24:41 -05:00
Andrei Betlen
4b0e3320bd fix: minor formatting bugs for chatml-function-calling 2024-02-13 03:11:35 -05:00
Andrei Betlen
6fe8b427e1 Bump version 2024-02-13 02:46:52 -05:00
Andrei Betlen
d1822fed6b fix: Don't change order of json schema object properties unless prop_order is passed, Closes #1180 2024-02-13 02:44:00 -05:00
Andrei Betlen
d605875772 Bump version 2024-02-12 16:28:30 -05:00
Andrei Betlen
cb791716b4 fix: Always set logits_all = True when using speculative decoding 2024-02-12 16:19:05 -05:00
Andrei
153a0049d9
feat: Generic chatml Function Calling (#957)
* Add demo notebook

* Add initial chat handler

* Update OpenAI types

* Add generic chatml function calling (wip)

* Update chatml generic function calling.

* Progress on auto-tool calls

* fix streaming functions

* Remove print statements

* fix: Suppress output from llama.cpp init and grammar creation

* Add OpenAI v1 python api compatible chat completion function

* Support non-streaming multi-tool calls

* Format

* Include function_call in response.
2024-02-12 15:56:07 -05:00
Andrei Betlen
69413ce08e Update llama.cpp 2024-02-11 19:00:17 -05:00
Connor
a05d90446f
fix: Circular dependancy preventing early Llama object free (#1176)
commit 901827013b introduced a cyclic dependency
within Llama objects. That change causes old models to linger in memory longer
than necessary, thereby creating memory bloat in most applications attempting
to switch between models at runtime. This patch simply removes the problematic
line, allowing models to deallocate without relying on GC. One might also
consider combining `weakref.ref` with a `@property` if the `llama` attribute is
absolutely necessary to expose in the tokenizer class.
2024-02-11 13:57:57 -05:00
Andrei Betlen
4abb8c9386 Merge branch 'main' of github.com:abetlen/llama_cpp_python into main 2024-02-09 13:32:31 -05:00
Andrei Betlen
e16f06e6eb fix: revert _create_completions. 2024-02-09 02:02:13 -05:00