Merge remote-tracking branch 'abetlen/main'

Author: oobabooga

CHANGELOG.md

@@ -7,6 +7,15 @@ and this project adheres to [Semantic Versioning](https://semver.org/spec/v2.0.0
 
 ## [Unreleased]
 
+## [0.2.83]
+
+- feat: Update llama.cpp to ggerganov/llama.cpp@081fe431aa8fb6307145c4feb3eed4f48cab19f8
+- feat: Add 'required' literal to ChatCompletionToolChoiceOption by @mjschock in #1597
+- fix: Change repeat_penalty to 1.0 to match llama.cpp defaults by @ddh0 in #1590
+- fix(docs): Update README.md typo by @ericcurtin in #1589
+- fix(server): Use split_mode from model settings by @grider-withourai in #1594
+- feat(ci): Dockerfile update base images and post-install cleanup by @Smartappli in #1530
+
 ## [0.2.82]
 
 - feat: Update llama.cpp to ggerganov/llama.cpp@7fdb6f73e35605c8dbc39e9f19cd9ed84dbc87f2

llama_cpp/__init__.py

@@ -1,4 +1,4 @@
 from .llama_cpp import *
 from .llama import *
 
-__version__ = "0.2.82.3"
+__version__ = "0.2.83"

llama_cpp/llama_cpp.py

@@ -247,8 +247,8 @@ def byref(obj: CtypesCData, offset: Optional[int] = None) -> CtypesRef[CtypesCDa
 
 # define LLAMA_SESSION_MAGIC   LLAMA_FILE_MAGIC_GGSN
 LLAMA_SESSION_MAGIC = LLAMA_FILE_MAGIC_GGSN
-# define LLAMA_SESSION_VERSION 6
-LLAMA_SESSION_VERSION = 6
+# define LLAMA_SESSION_VERSION 7
+LLAMA_SESSION_VERSION = 7
 
 # define LLAMA_STATE_SEQ_MAGIC   LLAMA_FILE_MAGIC_GGSQ
 LLAMA_STATE_SEQ_MAGIC = LLAMA_FILE_MAGIC_GGSQ
@@ -314,6 +314,9 @@ def byref(obj: CtypesCData, offset: Optional[int] = None) -> CtypesRef[CtypesCDa
 #     LLAMA_VOCAB_PRE_TYPE_CHATGLM4       = 17,
 #     LLAMA_VOCAB_PRE_TYPE_VIKING         = 18,
 #     LLAMA_VOCAB_PRE_TYPE_JAIS           = 19,
+#     LLAMA_VOCAB_PRE_TYPE_TEKKEN         = 20,
+#     LLAMA_VOCAB_PRE_TYPE_SMOLLM         = 21,
+#     LLAMA_VOCAB_PRE_TYPE_CODESHELL      = 22,
 # };
 LLAMA_VOCAB_PRE_TYPE_DEFAULT = 0
 LLAMA_VOCAB_PRE_TYPE_LLAMA3 = 1
@@ -335,6 +338,9 @@ def byref(obj: CtypesCData, offset: Optional[int] = None) -> CtypesRef[CtypesCDa
 LLAMA_VOCAB_PRE_TYPE_CHATGLM4 = 17
 LLAMA_VOCAB_PRE_TYPE_VIKING = 18
 LLAMA_VOCAB_PRE_TYPE_JAIS = 19
+LLAMA_VOCAB_PRE_TYPE_TEKKEN = 20
+LLAMA_VOCAB_PRE_TYPE_SMOLLM = 21
+LLAMA_VOCAB_PRE_TYPE_CODESHELL = 22
 
 
 # // note: these values should be synchronized with ggml_rope