llama-cpp-python 0.2.87__tar.gz → 0.2.88__tar.gz

llama_cpp_python-0.2.88/.git/FETCH_HEAD

@@ -0,0 +1 @@
+fc19cc783802ed95eb4c8be358537b12c1fec78f		'fc19cc783802ed95eb4c8be358537b12c1fec78f' of https://github.com/abetlen/llama-cpp-python

llama_cpp_python-0.2.88/.git/HEAD

@@ -0,0 +1 @@
+fc19cc783802ed95eb4c8be358537b12c1fec78f

{llama_cpp_python-0.2.87 → llama_cpp_python-0.2.88}/.git/config

@@ -9,7 +9,7 @@
 [gc]
 	auto = 0
 [http "https://github.com/"]
-	extraheader = AUTHORIZATION: basic eC1hY2Nlc3MtdG9rZW46Z2hzX3NGN2c0OFQ3cnlXRnI4Ujg4MXJMeENDUmdYQU01RjJmSDJ6Ng==
+	extraheader = AUTHORIZATION: basic eC1hY2Nlc3MtdG9rZW46Z2hzX2lnNnhqSEY0ZktIOTc2c0Z5U3IwUElQVGJSdnhhVjI3WEdFMQ==
 [submodule "vendor/llama.cpp"]
 	active = true
 	url = https://github.com/ggerganov/llama.cpp.git

llama_cpp_python-0.2.88/.git/logs/HEAD

@@ -0,0 +1 @@
+0000000000000000000000000000000000000000 fc19cc783802ed95eb4c8be358537b12c1fec78f runner <runner@fv-az774-340.j3qnnsaxsfoepktz21rketxmog.bx.internal.cloudapp.net> 1723540987 +0000	checkout: moving from master to refs/tags/v0.2.88

llama_cpp_python-0.2.88/.git/modules/vendor/llama.cpp/HEAD

@@ -0,0 +1 @@
+fc4ca27b25464a11b3b86c9dbb5b6ed6065965c2

{llama_cpp_python-0.2.87 → llama_cpp_python-0.2.88}/.git/modules/vendor/llama.cpp/config

@@ -16,7 +16,7 @@
 [gc]
 	auto = 0
 [http "https://github.com/"]
-	extraheader = AUTHORIZATION: basic eC1hY2Nlc3MtdG9rZW46Z2hzX3NGN2c0OFQ3cnlXRnI4Ujg4MXJMeENDUmdYQU01RjJmSDJ6Ng==
+	extraheader = AUTHORIZATION: basic eC1hY2Nlc3MtdG9rZW46Z2hzX2lnNnhqSEY0ZktIOTc2c0Z5U3IwUElQVGJSdnhhVjI3WEdFMQ==
 [url "https://github.com/"]
 	insteadOf = git@github.com:
 	insteadOf = org-6826477@github.com:

llama_cpp_python-0.2.88/.git/modules/vendor/llama.cpp/logs/HEAD

@@ -0,0 +1,2 @@
+0000000000000000000000000000000000000000 fc4ca27b25464a11b3b86c9dbb5b6ed6065965c2 runner <runner@fv-az774-340.j3qnnsaxsfoepktz21rketxmog.bx.internal.cloudapp.net> 1723540988 +0000	clone: from https://github.com/ggerganov/llama.cpp.git
+fc4ca27b25464a11b3b86c9dbb5b6ed6065965c2 fc4ca27b25464a11b3b86c9dbb5b6ed6065965c2 runner <runner@fv-az774-340.j3qnnsaxsfoepktz21rketxmog.bx.internal.cloudapp.net> 1723540988 +0000	checkout: moving from master to fc4ca27b25464a11b3b86c9dbb5b6ed6065965c2

llama_cpp_python-0.2.88/.git/modules/vendor/llama.cpp/logs/refs/heads/master

@@ -0,0 +1 @@
+0000000000000000000000000000000000000000 fc4ca27b25464a11b3b86c9dbb5b6ed6065965c2 runner <runner@fv-az774-340.j3qnnsaxsfoepktz21rketxmog.bx.internal.cloudapp.net> 1723540988 +0000	clone: from https://github.com/ggerganov/llama.cpp.git

llama_cpp_python-0.2.88/.git/modules/vendor/llama.cpp/logs/refs/remotes/origin/HEAD

@@ -0,0 +1 @@
+0000000000000000000000000000000000000000 fc4ca27b25464a11b3b86c9dbb5b6ed6065965c2 runner <runner@fv-az774-340.j3qnnsaxsfoepktz21rketxmog.bx.internal.cloudapp.net> 1723540988 +0000	clone: from https://github.com/ggerganov/llama.cpp.git

{llama_cpp_python-0.2.87 → llama_cpp_python-0.2.88}/.git/modules/vendor/llama.cpp/modules/kompute/config

@@ -13,7 +13,7 @@
 [gc]
 	auto = 0
 [http "https://github.com/"]
-	extraheader = AUTHORIZATION: basic eC1hY2Nlc3MtdG9rZW46Z2hzX3NGN2c0OFQ3cnlXRnI4Ujg4MXJMeENDUmdYQU01RjJmSDJ6Ng==
+	extraheader = AUTHORIZATION: basic eC1hY2Nlc3MtdG9rZW46Z2hzX2lnNnhqSEY0ZktIOTc2c0Z5U3IwUElQVGJSdnhhVjI3WEdFMQ==
 [url "https://github.com/"]
 	insteadOf = git@github.com:
 	insteadOf = org-6826477@github.com:

llama_cpp_python-0.2.88/.git/modules/vendor/llama.cpp/modules/kompute/logs/HEAD

@@ -0,0 +1,2 @@
+0000000000000000000000000000000000000000 f592b5bca3cbc169feb194218a086b18d618cca4 runner <runner@fv-az774-340.j3qnnsaxsfoepktz21rketxmog.bx.internal.cloudapp.net> 1723540988 +0000	clone: from https://github.com/nomic-ai/kompute.git
+f592b5bca3cbc169feb194218a086b18d618cca4 4565194ed7c32d1d2efa32ceab4d3c6cae006306 runner <runner@fv-az774-340.j3qnnsaxsfoepktz21rketxmog.bx.internal.cloudapp.net> 1723540989 +0000	checkout: moving from master to 4565194ed7c32d1d2efa32ceab4d3c6cae006306

llama_cpp_python-0.2.88/.git/modules/vendor/llama.cpp/modules/kompute/logs/refs/heads/master

@@ -0,0 +1 @@
+0000000000000000000000000000000000000000 f592b5bca3cbc169feb194218a086b18d618cca4 runner <runner@fv-az774-340.j3qnnsaxsfoepktz21rketxmog.bx.internal.cloudapp.net> 1723540988 +0000	clone: from https://github.com/nomic-ai/kompute.git

llama_cpp_python-0.2.88/.git/modules/vendor/llama.cpp/modules/kompute/logs/refs/remotes/origin/HEAD

@@ -0,0 +1 @@
+0000000000000000000000000000000000000000 f592b5bca3cbc169feb194218a086b18d618cca4 runner <runner@fv-az774-340.j3qnnsaxsfoepktz21rketxmog.bx.internal.cloudapp.net> 1723540988 +0000	clone: from https://github.com/nomic-ai/kompute.git

llama_cpp_python-0.2.88/.git/modules/vendor/llama.cpp/packed-refs

@@ -0,0 +1,2 @@
+# pack-refs with: peeled fully-peeled sorted 
+fc4ca27b25464a11b3b86c9dbb5b6ed6065965c2 refs/remotes/origin/master

llama_cpp_python-0.2.88/.git/modules/vendor/llama.cpp/refs/heads/master

@@ -0,0 +1 @@
+fc4ca27b25464a11b3b86c9dbb5b6ed6065965c2

llama_cpp_python-0.2.88/.git/modules/vendor/llama.cpp/shallow

@@ -0,0 +1 @@
+fc4ca27b25464a11b3b86c9dbb5b6ed6065965c2

llama_cpp_python-0.2.88/.git/refs/tags/v0.2.88

@@ -0,0 +1 @@
+fc19cc783802ed95eb4c8be358537b12c1fec78f

llama_cpp_python-0.2.88/.git/shallow

@@ -0,0 +1 @@
+fc19cc783802ed95eb4c8be358537b12c1fec78f

{llama_cpp_python-0.2.87 → llama_cpp_python-0.2.88}/.github/workflows/build-and-release.yaml

@@ -29,7 +29,7 @@ jobs:
           python -m pip install -e .[all]
 
       - name: Build wheels
-        uses: pypa/cibuildwheel@v2.19.2
+        uses: pypa/cibuildwheel@v2.20.0
         env:
           # disable repair
           CIBW_REPAIR_WHEEL_COMMAND: ""
@@ -56,7 +56,7 @@ jobs:
           platforms: linux/arm64
 
       - name: Build wheels
-        uses: pypa/cibuildwheel@v2.19.2
+        uses: pypa/cibuildwheel@v2.20.0
         env:
           CIBW_SKIP: "*musllinux* pp*"
           CIBW_REPAIR_WHEEL_COMMAND: ""

{llama_cpp_python-0.2.87 → llama_cpp_python-0.2.88}/.github/workflows/build-wheels-metal.yaml

@@ -30,7 +30,7 @@ jobs:
           python -m pip install -e .[all]
 
       - name: Build wheels
-        uses: pypa/cibuildwheel@v2.19.2
+        uses: pypa/cibuildwheel@v2.20.0
         env:
           # disable repair
           CIBW_REPAIR_WHEEL_COMMAND: ""

{llama_cpp_python-0.2.87 → llama_cpp_python-0.2.88}/CHANGELOG.md

@@ -7,6 +7,15 @@ and this project adheres to [Semantic Versioning](https://semver.org/spec/v2.0.0
 
 ## [Unreleased]
 
+## [0.2.88]
+
+- feat: Update llama.cpp to ggerganov/llama.cpp@fc4ca27b25464a11b3b86c9dbb5b6ed6065965c2
+- fix: only print 'cache saved' in verbose mode by @lsorber in #1668 
+- fix: Added back from_file method to LlamaGrammar by @ExtReMLapin in #1673
+- fix: grammar prints on each call by @abetlen in 0998ea0deea076a547d54bd598d6b413b588ee2b
+- feat: Enable recursive search of HFFS.ls when using from_pretrained by @benHeidabetlen in #1656
+- feat: Add more detailed log for prefix-match by @xu-song in #1659
+
 ## [0.2.87]
 
 - feat: Update llama.cpp to ggerganov/llama.cpp@be55695eff44784a141a863f273661a6bce63dfc

{llama_cpp_python-0.2.87 → llama_cpp_python-0.2.88}/PKG-INFO

@@ -1,6 +1,6 @@
 Metadata-Version: 2.1
 Name: llama_cpp_python
-Version: 0.2.87
+Version: 0.2.88
 Summary: Python bindings for the llama.cpp library
 Author-Email: Andrei Betlen <abetlen@gmail.com>
 License: MIT

{llama_cpp_python-0.2.87 → llama_cpp_python-0.2.88}/llama_cpp/__init__.py

@@ -1,4 +1,4 @@
 from .llama_cpp import *
 from .llama import *
 
-__version__ = "0.2.87"
+__version__ = "0.2.88"

{llama_cpp_python-0.2.87 → llama_cpp_python-0.2.88}/llama_cpp/llama.py

@@ -777,11 +777,12 @@ class Llama:
                 else:
                     break
             if longest_prefix > 0:
-                if self.verbose:
-                    print("Llama.generate: prefix-match hit", file=sys.stderr)
                 reset = False
                 tokens = tokens[longest_prefix:]
                 self.n_tokens = longest_prefix
+                if self.verbose:
+                    print(f"Llama.generate: {longest_prefix} prefix-match hit, "
+                          f"remaining {len(tokens)} prompt tokens to eval", file=sys.stderr)                    
 
         # Reset the model state
         if reset:
@@ -1522,7 +1523,8 @@ class Llama:
                 if self.verbose:
                     print("Llama._create_completion: cache save", file=sys.stderr)
                 self.cache[prompt_tokens + completion_tokens] = self.save_state()
-                print("Llama._create_completion: cache saved", file=sys.stderr)
+                if self.verbose:
+                    print("Llama._create_completion: cache saved", file=sys.stderr)
             return
 
         if self.cache:
@@ -2159,7 +2161,7 @@ class Llama:
 
         files = [
             file["name"] if isinstance(file, dict) else file
-            for file in hffs.ls(repo_id)
+            for file in hffs.ls(repo_id, recursive=True)
         ]
 
         # split each file into repo_id, subfolder, filename

{llama_cpp_python-0.2.87 → llama_cpp_python-0.2.88}/llama_cpp/llama_cpp.py

@@ -1505,6 +1505,14 @@ def llama_model_has_encoder(model: llama_model_p, /) -> bool:
     ...
 
 
+# // Returns true if the model contains a decoder that requires llama_decode() call
+# LLAMA_API bool llama_model_has_decoder(const struct llama_model * model);
+@ctypes_function("llama_model_has_decoder", [llama_model_p_ctypes], ctypes.c_bool)
+def llama_model_has_decoder(model: llama_model_p, /) -> bool:
+    """Returns true if the model contains a decoder that requires llama_decode() call"""
+    ...
+
+
 # // For encoder-decoder models, this function returns id of the token that must be provided
 # // to the decoder to start generating output sequence. For other models, it returns -1.
 # LLAMA_API llama_token llama_model_decoder_start_token(const struct llama_model * model);

{llama_cpp_python-0.2.87 → llama_cpp_python-0.2.88}/llama_cpp/llama_grammar.py

@@ -1,6 +1,7 @@
 """Python implementation of llama grammar parser directly translated from C++ source file in vendor/llama.cpp/common/grammar-parser.cpp."""
 
 # flake8: noqa
+from pathlib import Path
 import sys
 import ctypes
 import enum
@@ -890,8 +891,26 @@ class LlamaGrammar:
     @classmethod
     def from_string(cls, grammar: str, verbose: bool = True) -> "LlamaGrammar":
         parsed_grammar = parse(grammar)
-        print_grammar(file=sys.stdout, state=parsed_grammar)
+        if verbose:
+            print_grammar(file=sys.stdout, state=parsed_grammar)
         return cls(parsed_grammar)
+    
+    @classmethod
+    def from_file(cls, file: Union[str, Path], verbose: bool = True) -> "LlamaGrammar":
+        try:
+            with open(file) as f:
+                grammar = f.read()
+        except Exception as err:
+            raise Exception(
+                f"{cls.from_file.__name__}: error reading grammar file: {err}"
+            )
+
+        if grammar:
+            return cls.from_string(grammar, verbose=verbose)
+
+        raise ValueError(
+            f"{cls.from_file.__name__}: error parsing grammar file: params_grammer is empty"
+        )
 
     @classmethod
     def from_json_schema(cls, json_schema: str, verbose: bool = True) -> "LlamaGrammar":

{llama_cpp_python-0.2.87 → llama_cpp_python-0.2.88}/vendor/llama.cpp/.github/workflows/bench.yml

@@ -129,6 +129,8 @@ jobs:
 
       - name: Server bench
         id: server_bench
+        env:
+            HEAD_REF: ${{ github.head_ref || github.ref_name }}
         run: |
           set -eux
 
@@ -137,7 +139,7 @@ jobs:
           python bench.py \
               --runner-label ${{ env.RUNNER_LABEL }} \
               --name ${{ github.job }} \
-              --branch ${{ github.head_ref || github.ref_name }} \
+              --branch $HEAD_REF \
               --commit ${{ github.event.inputs.sha || github.event.pull_request.head.sha || github.sha }} \
               --scenario script.js \
               --duration ${{ github.event.inputs.duration || env.DURATION }} \

{llama_cpp_python-0.2.87 → llama_cpp_python-0.2.88}/vendor/llama.cpp/.github/workflows/build.yml

@@ -47,7 +47,7 @@ jobs:
           sysctl -a
           mkdir build
           cd build
-          cmake -DLLAMA_FATAL_WARNINGS=ON -DGGML_METAL_EMBED_LIBRARY=ON -DLLAMA_CURL=ON -DBUILD_SHARED_LIBS=OFF ..
+          cmake -DLLAMA_FATAL_WARNINGS=ON -DGGML_METAL_EMBED_LIBRARY=ON -DLLAMA_CURL=ON -DGGML_RPC=ON -DBUILD_SHARED_LIBS=OFF ..
           cmake --build . --config Release -j $(sysctl -n hw.logicalcpu)
 
       - name: Test
@@ -105,7 +105,7 @@ jobs:
           sysctl -a
           # Metal is disabled due to intermittent failures with Github runners not having a GPU:
           # https://github.com/ggerganov/llama.cpp/actions/runs/8635935781/job/23674807267#step:5:2313
-          cmake -B build -DLLAMA_FATAL_WARNINGS=ON -DGGML_METAL=OFF -DLLAMA_CURL=ON -DBUILD_SHARED_LIBS=OFF
+          cmake -B build -DLLAMA_FATAL_WARNINGS=ON -DGGML_METAL=OFF -DLLAMA_CURL=ON -DGGML_RPC=ON -DBUILD_SHARED_LIBS=OFF
           cmake --build build --config Release -j $(sysctl -n hw.logicalcpu)
 
       - name: Test
@@ -222,7 +222,7 @@ jobs:
         run: |
           mkdir build
           cd build
-          cmake .. -DLLAMA_FATAL_WARNINGS=ON -DLLAMA_CURL=ON -DBUILD_SHARED_LIBS=OFF
+          cmake .. -DLLAMA_FATAL_WARNINGS=ON -DLLAMA_CURL=ON -DGGML_RPC=ON -DBUILD_SHARED_LIBS=OFF
           cmake --build . --config Release -j $(nproc)
 
       - name: Test
@@ -696,22 +696,20 @@ jobs:
     strategy:
       matrix:
         include:
-          - build: 'rpc-x64'
-            defines: '-DGGML_NATIVE=OFF -DLLAMA_BUILD_SERVER=ON -DGGML_RPC=ON -DBUILD_SHARED_LIBS=ON'
           - build: 'noavx-x64'
-            defines: '-DGGML_NATIVE=OFF -DLLAMA_BUILD_SERVER=ON -DGGML_AVX=OFF -DGGML_AVX2=OFF -DGGML_FMA=OFF -DBUILD_SHARED_LIBS=ON'
+            defines: '-DGGML_NATIVE=OFF -DLLAMA_BUILD_SERVER=ON -DGGML_RPC=ON -DGGML_AVX=OFF -DGGML_AVX2=OFF -DGGML_FMA=OFF -DBUILD_SHARED_LIBS=ON'
           - build: 'avx2-x64'
-            defines: '-DGGML_NATIVE=OFF -DLLAMA_BUILD_SERVER=ON -DBUILD_SHARED_LIBS=ON'
+            defines: '-DGGML_NATIVE=OFF -DLLAMA_BUILD_SERVER=ON -DGGML_RPC=ON -DBUILD_SHARED_LIBS=ON'
           - build: 'avx-x64'
-            defines: '-DGGML_NATIVE=OFF -DLLAMA_BUILD_SERVER=ON -DGGML_AVX2=OFF -DBUILD_SHARED_LIBS=ON'
+            defines: '-DGGML_NATIVE=OFF -DLLAMA_BUILD_SERVER=ON -DGGML_RPC=ON -DGGML_AVX2=OFF -DBUILD_SHARED_LIBS=ON'
           - build: 'avx512-x64'
-            defines: '-DGGML_NATIVE=OFF -DLLAMA_BUILD_SERVER=ON -DGGML_AVX512=ON -DBUILD_SHARED_LIBS=ON'
+            defines: '-DGGML_NATIVE=OFF -DLLAMA_BUILD_SERVER=ON -DGGML_RPC=ON -DGGML_AVX512=ON -DBUILD_SHARED_LIBS=ON'
           - build: 'openblas-x64'
-            defines: '-DGGML_NATIVE=OFF -DLLAMA_BUILD_SERVER=ON -DGGML_BLAS=ON -DBUILD_SHARED_LIBS=ON -DGGML_BLAS_VENDOR=OpenBLAS -DBLAS_INCLUDE_DIRS="$env:RUNNER_TEMP/openblas/include" -DBLAS_LIBRARIES="$env:RUNNER_TEMP/openblas/lib/openblas.lib"'
+            defines: '-DGGML_NATIVE=OFF -DLLAMA_BUILD_SERVER=ON -DGGML_RPC=ON -DGGML_BLAS=ON -DBUILD_SHARED_LIBS=ON -DGGML_BLAS_VENDOR=OpenBLAS -DBLAS_INCLUDE_DIRS="$env:RUNNER_TEMP/openblas/include" -DBLAS_LIBRARIES="$env:RUNNER_TEMP/openblas/lib/openblas.lib"'
           - build: 'kompute-x64'
-            defines: '-DGGML_NATIVE=OFF -DLLAMA_BUILD_SERVER=ON -DGGML_KOMPUTE=ON -DKOMPUTE_OPT_DISABLE_VULKAN_VERSION_CHECK=ON -DBUILD_SHARED_LIBS=ON'
+            defines: '-DGGML_NATIVE=OFF -DLLAMA_BUILD_SERVER=ON -DGGML_RPC=ON -DGGML_KOMPUTE=ON -DKOMPUTE_OPT_DISABLE_VULKAN_VERSION_CHECK=ON -DBUILD_SHARED_LIBS=ON'
           - build: 'vulkan-x64'
-            defines: '-DGGML_NATIVE=OFF -DLLAMA_BUILD_SERVER=ON -DGGML_VULKAN=ON -DBUILD_SHARED_LIBS=ON'
+            defines: '-DGGML_NATIVE=OFF -DLLAMA_BUILD_SERVER=ON -DGGML_RPC=ON -DGGML_VULKAN=ON -DBUILD_SHARED_LIBS=ON'
           - build: 'llvm-arm64'
             defines: '-G "Ninja Multi-Config" -D CMAKE_TOOLCHAIN_FILE=cmake/arm64-windows-llvm.cmake -DGGML_NATIVE=OFF -DLLAMA_BUILD_SERVER=ON -DBUILD_SHARED_LIBS=ON'
           - build: 'msvc-arm64'

{llama_cpp_python-0.2.87 → llama_cpp_python-0.2.88}/vendor/llama.cpp/.github/workflows/python-check-requirements.yml

@@ -6,15 +6,13 @@ on:
       - '.github/workflows/python-check-requirements.yml'
       - 'scripts/check-requirements.sh'
       - 'convert*.py'
-      - 'requirements.txt'
-      - 'requirements/*.txt'
+      - '**/requirements*.txt'
   pull_request:
     paths:
       - '.github/workflows/python-check-requirements.yml'
       - 'scripts/check-requirements.sh'
       - 'convert*.py'
-      - 'requirements.txt'
-      - 'requirements/*.txt'
+      - '**/requirements*.txt'
 
 concurrency:
   group: ${{ github.workflow }}-${{ github.head_ref && github.ref || github.run_id }}

{llama_cpp_python-0.2.87 → llama_cpp_python-0.2.88}/vendor/llama.cpp/.gitignore

@@ -79,7 +79,6 @@ models-mnt
 !models/ggml-vocab-*.gguf*
 
 # Zig
-
 zig-out/
 zig-cache/
 

{llama_cpp_python-0.2.87 → llama_cpp_python-0.2.88}/vendor/llama.cpp/Makefile

@@ -19,6 +19,7 @@ BUILD_TARGETS = \
 	llama-imatrix \
 	llama-infill \
 	llama-llava-cli \
+	llama-minicpmv-cli\
 	llama-lookahead \
 	llama-lookup \
 	llama-lookup-create \
@@ -888,15 +889,16 @@ ggml/src/ggml-metal-embed.o: \
 	ggml/src/ggml-common.h
 	@echo "Embedding Metal library"
 	@sed -e '/#include "ggml-common.h"/r ggml/src/ggml-common.h' -e '/#include "ggml-common.h"/d' < ggml/src/ggml-metal.metal > ggml/src/ggml-metal-embed.metal
-	$(eval TEMP_ASSEMBLY=$(shell mktemp))
-	@echo ".section __DATA, __ggml_metallib"            >  $(TEMP_ASSEMBLY)
-	@echo ".globl _ggml_metallib_start"                 >> $(TEMP_ASSEMBLY)
-	@echo "_ggml_metallib_start:"                       >> $(TEMP_ASSEMBLY)
-	@echo ".incbin \"ggml/src/ggml-metal-embed.metal\"" >> $(TEMP_ASSEMBLY)
-	@echo ".globl _ggml_metallib_end"                   >> $(TEMP_ASSEMBLY)
-	@echo "_ggml_metallib_end:"                         >> $(TEMP_ASSEMBLY)
-	@$(AS) $(TEMP_ASSEMBLY) -o $@
-	@rm -f ${TEMP_ASSEMBLY}
+	$(eval TEMP_ASSEMBLY=$(shell mktemp -d))
+	@echo ".section __DATA, __ggml_metallib"            >  $(TEMP_ASSEMBLY)/ggml-metal-embed.s
+	@echo ".globl _ggml_metallib_start"                 >> $(TEMP_ASSEMBLY)/ggml-metal-embed.s
+	@echo "_ggml_metallib_start:"                       >> $(TEMP_ASSEMBLY)/ggml-metal-embed.s
+	@echo ".incbin \"ggml/src/ggml-metal-embed.metal\"" >> $(TEMP_ASSEMBLY)/ggml-metal-embed.s
+	@echo ".globl _ggml_metallib_end"                   >> $(TEMP_ASSEMBLY)/ggml-metal-embed.s
+	@echo "_ggml_metallib_end:"                         >> $(TEMP_ASSEMBLY)/ggml-metal-embed.s
+	$(CC) $(CFLAGS) -c $(TEMP_ASSEMBLY)/ggml-metal-embed.s -o $@
+	@rm -f ${TEMP_ASSEMBLY}/ggml-metal-embed.s
+	@rmdir ${TEMP_ASSEMBLY}
 endif
 endif # GGML_METAL
 
@@ -1205,6 +1207,7 @@ clean:
 	rm -rvf ggml/*.dll
 	rm -rvf ggml/*.so
 	rm -vrf ggml/src/*.o
+	rm -rvf ggml/src/llamafile/*.o
 	rm -rvf common/build-info.cpp
 	rm -vrf ggml/src/ggml-metal-embed.metal
 	rm -vrf ggml/src/ggml-cuda/*.o
@@ -1451,15 +1454,20 @@ libllava.a: examples/llava/llava.cpp \
 	$(CXX) $(CXXFLAGS) -static -fPIC -c $< -o $@ -Wno-cast-qual
 
 llama-llava-cli: examples/llava/llava-cli.cpp \
-	examples/llava/clip.h \
-	examples/llava/clip.cpp \
+	examples/llava/llava.cpp \
 	examples/llava/llava.h \
+	examples/llava/clip.cpp \
+	examples/llava/clip.h \
+	$(OBJ_ALL)
+	$(CXX) $(CXXFLAGS) $< $(filter-out %.h $<,$^) -o $@ $(LDFLAGS) -Wno-cast-qual
+
+llama-minicpmv-cli: examples/llava/minicpmv-cli.cpp \
 	examples/llava/llava.cpp \
+	examples/llava/llava.h \
+	examples/llava/clip.cpp \
+	examples/llava/clip.h \
 	$(OBJ_ALL)
-	$(CXX) $(CXXFLAGS) -c $< -o $(call GET_OBJ_FILE, $<)
-	$(CXX) $(CXXFLAGS) -c examples/llava/clip.cpp  -o $(call GET_OBJ_FILE, examples/llava/clip.cpp) -Wno-cast-qual
-	$(CXX) $(CXXFLAGS) -c examples/llava/llava.cpp -o $(call GET_OBJ_FILE, examples/llava/llava.cpp)
-	$(CXX) $(CXXFLAGS) $(filter-out %.h $< examples/llava/clip.cpp examples/llava/llava.cpp,$^) $(call GET_OBJ_FILE, $<) $(call GET_OBJ_FILE, examples/llava/clip.cpp) $(call GET_OBJ_FILE, examples/llava/llava.cpp) -o $@ $(LDFLAGS)
+	$(CXX) $(CXXFLAGS) $< $(filter-out %.h $<,$^) -o $@ $(LDFLAGS) -Wno-cast-qual
 
 ifeq ($(UNAME_S),Darwin)
 swift: examples/batched.swift

{llama_cpp_python-0.2.87 → llama_cpp_python-0.2.88}/vendor/llama.cpp/README.md

@@ -186,10 +186,12 @@ Unless otherwise noted these projects are open-source with permissive licensing:
 
 - [akx/ggify](https://github.com/akx/ggify) – download PyTorch models from HuggingFace Hub and convert them to GGML
 - [crashr/gppm](https://github.com/crashr/gppm) – launch llama.cpp instances utilizing NVIDIA Tesla P40 or P100 GPUs with reduced idle power consumption
+- [gpustack/gguf-parser](https://github.com/gpustack/gguf-parser-go/tree/main/cmd/gguf-parser) - review/check the GGUF file and estimate the memory usage
 
 **Infrastructure:**
 
 - [Paddler](https://github.com/distantmagic/paddler) - Stateful load balancer custom-tailored for llama.cpp
+- [GPUStack](https://github.com/gpustack/gpustack) - Manage GPU clusters for running LLMs
 
 **Games:**
 - [Lucy's Labyrinth](https://github.com/MorganRO8/Lucys_Labyrinth) - A simple maze game where agents controlled by an AI model will try to trick you.

{llama_cpp_python-0.2.87 → llama_cpp_python-0.2.88}/vendor/llama.cpp/common/build-info.cpp

@@ -1,4 +1,4 @@
 int LLAMA_BUILD_NUMBER = 1;
-char const *LLAMA_COMMIT = "be55695";
+char const *LLAMA_COMMIT = "fc4ca27";
 char const *LLAMA_COMPILER = "gcc (Ubuntu 11.4.0-1ubuntu1~22.04) 11.4.0";
 char const *LLAMA_BUILD_TARGET = "x86_64-linux-gnu";

{llama_cpp_python-0.2.87 → llama_cpp_python-0.2.88}/vendor/llama.cpp/common/common.cpp

@@ -1777,6 +1777,17 @@ std::string string_get_sortable_timestamp() {
     return std::string(timestamp_no_ns) + "." + std::string(timestamp_ns);
 }
 
+void string_replace_all(std::string & s, const std::string & search, const std::string & replace) {
+    if (search.empty()) {
+        return; // Avoid infinite loop if 'search' is an empty string
+    }
+    size_t pos = 0;
+    while ((pos = s.find(search, pos)) != std::string::npos) {
+        s.replace(pos, search.length(), replace);
+        pos += replace.length();
+    }
+}
+
 void string_process_escapes(std::string & input) {
     std::size_t input_len = input.length();
     std::size_t output_idx = 0;
@@ -2145,7 +2156,9 @@ struct llama_init_result llama_init_from_gpt_params(gpt_params & params) {
             tmp.clear();
             tmp.push_back(decoder_start_token_id);
         }
-        llama_decode(lctx, llama_batch_get_one(tmp.data(), std::min(tmp.size(), (size_t) params.n_batch), 0, 0));
+        if (llama_model_has_decoder(model)) {
+            llama_decode(lctx, llama_batch_get_one(tmp.data(), std::min(tmp.size(), (size_t) params.n_batch), 0, 0));
+        }
         llama_kv_cache_clear(lctx);
         llama_synchronize(lctx);
         llama_reset_timings(lctx);

{llama_cpp_python-0.2.87 → llama_cpp_python-0.2.88}/vendor/llama.cpp/common/common.h

@@ -286,6 +286,8 @@ std::vector<std::string> string_split(std::string input, char separator);
 std::string string_strip(const std::string & str);
 std::string string_get_sortable_timestamp();
 
+void string_replace_all(std::string & s, const std::string & search, const std::string & replace);
+
 template<class T>
 static std::vector<T> string_split(const std::string & str, char delim) {
     std::vector<T> values;

{llama_cpp_python-0.2.87 → llama_cpp_python-0.2.88}/vendor/llama.cpp/common/grammar-parser.cpp

@@ -369,6 +369,9 @@ namespace grammar_parser {
             }
             // Validate the state to ensure that all rules are defined
             for (const auto & rule : state.rules) {
+                if (rule.empty()) {
+                    throw std::runtime_error("Undefined rule");
+                }
                 for (const auto & elem : rule) {
                     if (elem.type == LLAMA_GRETYPE_RULE_REF) {
                         // Ensure that the rule at that location exists