bigdl-core-cpp 2.5.0b20240508__py3-none-win_amd64.whl → 2.5.0b20240510__py3-none-win_amd64.whl

bigdl/cpp/convert.py

@@ -1,6 +1,7 @@
 #!/usr/bin/env python3
 from __future__ import annotations
 
+import logging
 import argparse
 import concurrent.futures
 import enum
@@ -35,6 +36,8 @@ import gguf
 if TYPE_CHECKING:
     from typing_extensions import Self, TypeAlias
 
+logger = logging.getLogger("convert")
+
 if hasattr(faulthandler, 'register') and hasattr(signal, 'SIGUSR1'):
     faulthandler.register(signal.SIGUSR1)
 
@@ -643,7 +646,6 @@ class LlamaHfVocab(Vocab):
 
 
 def permute(weights: NDArray, n_head: int, n_head_kv: int) -> NDArray:
-    # print( "permute debug " + str(weights.shape[0]) + " x " + str(weights.shape[1]) + " nhead " + str(n_head) + " nheadkv " + str(n_kv_head) )
     if n_head_kv is not None and n_head != n_head_kv:
         n_head = n_head_kv
     return (weights.reshape(n_head, 2, weights.shape[0] // n_head // 2, *weights.shape[1:])
@@ -1033,12 +1035,12 @@ def check_vocab_size(params: Params, vocab: BaseVocab, pad_vocab: bool = False)
 
     # Check for a vocab size mismatch
     if params.n_vocab == vocab.vocab_size:
-        print("Ignoring added_tokens.json since model matches vocab size without it.")
+        logger.warning("Ignoring added_tokens.json since model matches vocab size without it.")
         return
 
     if pad_vocab and params.n_vocab > vocab.vocab_size:
         pad_count = params.n_vocab - vocab.vocab_size
-        print(
+        logger.debug(
             f"Padding vocab with {pad_count} token(s) - <dummy00001> through <dummy{pad_count:05}>"
         )
         for i in range(1, pad_count + 1):
@@ -1166,7 +1168,7 @@ class OutputFile:
             elapsed = time.time() - start
             size = ' x '.join(f"{dim:6d}" for dim in lazy_tensor.shape)
             padi = len(str(len(model)))
-            print(
+            logger.info(
                 f"[{i + 1:{padi}d}/{len(model)}] Writing tensor {name:38s} | size {size:16} | type {lazy_tensor.data_type.name:4} | T+{int(elapsed):4}"
             )
             self.gguf.write_tensor_data(ndarray)
@@ -1281,12 +1283,12 @@ def convert_model_names(model: LazyModel, params: Params, skip_unknown: bool) ->
     # HF models permut or pack some of the tensors, so we need to undo that
     for i in itertools.count():
         if f"model.layers.{i}.self_attn.q_proj.weight" in model:
-            print(f"Permuting layer {i}")
+            logger.debug(f"Permuting layer {i}")
             tmp[f"model.layers.{i}.self_attn.q_proj.weight"] = permute_lazy(model[f"model.layers.{i}.self_attn.q_proj.weight"], params.n_head, params.n_head)
             tmp[f"model.layers.{i}.self_attn.k_proj.weight"] = permute_lazy(model[f"model.layers.{i}.self_attn.k_proj.weight"], params.n_head, params.n_head_kv)
             # tmp[f"model.layers.{i}.self_attn.v_proj.weight"] =              model[f"model.layers.{i}.self_attn.v_proj.weight"]
         elif f"model.layers.{i}.self_attn.W_pack.weight" in model:
-            print(f"Unpacking and permuting layer {i}")
+            logger.debug(f"Unpacking and permuting layer {i}")
             tmp[f"model.layers.{i}.self_attn.q_proj.weight"] = permute_part_lazy(model[f"model.layers.{i}.self_attn.W_pack.weight"], 0, params.n_head, params.n_head)
             tmp[f"model.layers.{i}.self_attn.k_proj.weight"] = permute_part_lazy(model[f"model.layers.{i}.self_attn.W_pack.weight"], 1, params.n_head, params.n_head_kv)
             tmp[f"model.layers.{i}.self_attn.v_proj.weight"] = part_lazy        (model[f"model.layers.{i}.self_attn.W_pack.weight"], 2)
@@ -1299,15 +1301,15 @@ def convert_model_names(model: LazyModel, params: Params, skip_unknown: bool) ->
         tensor_type, name_new = tmap.get_type_and_name(name, try_suffixes = (".weight", ".bias")) or (None, None)
         if name_new is None:
             if skip_unknown:
-                print(f"Unexpected tensor name: {name} - skipping")
+                logger.warning(f"Unexpected tensor name: {name} - skipping")
                 continue
             raise ValueError(f"Unexpected tensor name: {name}. Use --skip-unknown to ignore it (e.g. LLaVA)")
 
         if tensor_type in should_skip:
-            print(f"skipping tensor {name_new}")
+            logger.debug(f"skipping tensor {name_new}")
             continue
 
-        print(f"{name:48s} -> {name_new:40s} | {lazy_tensor.data_type.name:6s} | {lazy_tensor.shape}")
+        logger.debug(f"{name:48s} -> {name_new:40s} | {lazy_tensor.data_type.name:6s} | {lazy_tensor.shape}")
         out[name_new] = lazy_tensor
 
     return out
@@ -1372,7 +1374,7 @@ def load_some_model(path: Path) -> ModelPlus:
     paths = find_multifile_paths(path)
     models_plus: list[ModelPlus] = []
     for path in paths:
-        print(f"Loading model file {path}")
+        logger.info(f"Loading model file {path}")
         models_plus.append(lazy_load_file(path))
 
     model_plus = merge_multifile_models(models_plus)
@@ -1413,7 +1415,7 @@ class VocabFactory:
         else:
             raise FileNotFoundError(f"Could not find a tokenizer matching any of {vocab_types}")
 
-        print(f"Loaded vocab file {vocab.fname_tokenizer!r}, type {vocab.name!r}")
+        logger.info(f"Loaded vocab file {vocab.fname_tokenizer!r}, type {vocab.name!r}")
         return vocab
 
     def load_vocab(self, vocab_types: list[str] | None, model_parent_path: Path) -> tuple[BaseVocab, gguf.SpecialVocab]:
@@ -1438,19 +1440,19 @@ def default_outfile(model_paths: list[Path], file_type: GGMLFileType) -> Path:
     }[file_type]
     ret = model_paths[0].parent / f"ggml-model-{namestr}.gguf"
     if ret in model_paths:
-        sys.stderr.write(
+        logger.error(
             f"Error: Default output path ({ret}) would overwrite the input. "
-            "Please explicitly specify a path using --outfile.\n")
+            "Please explicitly specify a path using --outfile.")
         sys.exit(1)
     return ret
 
 
 def do_dump_model(model_plus: ModelPlus) -> None:
-    print(f"model_plus.paths = {model_plus.paths!r}")
-    print(f"model_plus.format = {model_plus.format!r}")
-    print(f"model_plus.vocab = {model_plus.vocab!r}")
+    print(f"model_plus.paths = {model_plus.paths!r}") # noqa: NP100
+    print(f"model_plus.format = {model_plus.format!r}") # noqa: NP100
+    print(f"model_plus.vocab = {model_plus.vocab!r}") # noqa: NP100
     for name, lazy_tensor in model_plus.model.items():
-        print(f"{name}: shape={lazy_tensor.shape} type={lazy_tensor.data_type}; {lazy_tensor.description}")
+        print(f"{name}: shape={lazy_tensor.shape} type={lazy_tensor.data_type}; {lazy_tensor.description}") # noqa: NP100
 
 
 def main(args_in: list[str] | None = None) -> None:
@@ -1473,8 +1475,18 @@ def main(args_in: list[str] | None = None) -> None:
     parser.add_argument("--big-endian",   action="store_true",    help="model is executed on big endian machine")
     parser.add_argument("--pad-vocab",    action="store_true",    help="add pad tokens when model vocab expects more than tokenizer metadata provides")
     parser.add_argument("--skip-unknown", action="store_true",    help="skip unknown tensor names instead of failing")
+    parser.add_argument("--verbose",      action="store_true",    help="increase output verbosity")
 
     args = parser.parse_args(args_in)
+
+    if args.verbose:
+        logging.basicConfig(level=logging.DEBUG)
+    elif args.dump_single or args.dump:
+        # Avoid printing anything besides the dump output
+        logging.basicConfig(level=logging.WARNING)
+    else:
+        logging.basicConfig(level=logging.INFO)
+
     if args.no_vocab and args.vocab_only:
         raise ValueError("--vocab-only does not make sense with --no-vocab")
 
@@ -1491,6 +1503,7 @@ def main(args_in: list[str] | None = None) -> None:
     if args.dump:
         do_dump_model(model_plus)
         return
+
     endianess = gguf.GGUFEndian.LITTLE
     if args.big_endian:
         endianess = gguf.GGUFEndian.BIG
@@ -1513,7 +1526,7 @@ def main(args_in: list[str] | None = None) -> None:
             "q8_0": GGMLFileType.MostlyQ8_0,
         }[args.outtype]
 
-    print(f"params = {params}")
+    logger.info(f"params = {params}")
 
     model_parent_path = model_plus.paths[0].parent
     vocab_path = Path(args.vocab_dir or args.model or model_parent_path)
@@ -1528,15 +1541,14 @@ def main(args_in: list[str] | None = None) -> None:
         outfile = args.outfile
         OutputFile.write_vocab_only(outfile, params, vocab, special_vocab,
                                     endianess=endianess, pad_vocab=args.pad_vocab)
-        print(f"Wrote {outfile}")
+        logger.info(f"Wrote {outfile}")
         return
 
     if model_plus.vocab is not None and args.vocab_dir is None and not args.no_vocab:
         vocab = model_plus.vocab
 
-    print(f"Vocab info: {vocab}")
-    print(f"Special vocab info: {special_vocab}")
-
+    logger.info(f"Vocab info: {vocab}")
+    logger.info(f"Special vocab info: {special_vocab}")
     model   = model_plus.model
     model   = convert_model_names(model, params, args.skip_unknown)
     ftype   = pick_output_type(model, args.outtype)
@@ -1544,11 +1556,11 @@ def main(args_in: list[str] | None = None) -> None:
     outfile = args.outfile or default_outfile(model_plus.paths, ftype)
 
     params.ftype = ftype
-    print(f"Writing {outfile}, format {ftype}")
+    logger.info(f"Writing {outfile}, format {ftype}")
 
     OutputFile.write_all(outfile, ftype, params, model, vocab, special_vocab,
                          concurrency=args.concurrency, endianess=endianess, pad_vocab=args.pad_vocab)
-    print(f"Wrote {outfile}")
+    logger.info(f"Wrote {outfile}")
 
 
 if __name__ == '__main__':

bigdl/cpp/gguf-py/gguf/constants.py

@@ -1,6 +1,5 @@
 from __future__ import annotations
 
-import sys
 from enum import Enum, IntEnum, auto
 from typing import Any
 
@@ -72,6 +71,7 @@ class Keys:
 
     class Tokenizer:
         MODEL            = "tokenizer.ggml.model"
+        PRE              = "tokenizer.ggml.pre"
         LIST             = "tokenizer.ggml.tokens"
         TOKEN_TYPE       = "tokenizer.ggml.token_type"
         TOKEN_TYPE_COUNT = "tokenizer.ggml.token_type_count"  # for BERT-style token types
@@ -124,6 +124,7 @@ class MODEL_ARCH(IntEnum):
     QWEN2      = auto()
     QWEN2MOE   = auto()
     PHI2       = auto()
+    PHI3       = auto()
     PLAMO      = auto()
     CODESHELL  = auto()
     ORION      = auto()
@@ -200,6 +201,7 @@ MODEL_ARCH_NAMES: dict[MODEL_ARCH, str] = {
     MODEL_ARCH.QWEN2:          "qwen2",
     MODEL_ARCH.QWEN2MOE:       "qwen2moe",
     MODEL_ARCH.PHI2:           "phi2",
+    MODEL_ARCH.PHI3:           "phi3",
     MODEL_ARCH.PLAMO:          "plamo",
     MODEL_ARCH.CODESHELL:      "codeshell",
     MODEL_ARCH.ORION:          "orion",
@@ -550,6 +552,20 @@ MODEL_TENSORS: dict[MODEL_ARCH, list[MODEL_TENSOR]] = {
         MODEL_TENSOR.FFN_DOWN,
         MODEL_TENSOR.FFN_UP,
     ],
+    MODEL_ARCH.PHI3: [
+        MODEL_TENSOR.TOKEN_EMBD,
+        MODEL_TENSOR.OUTPUT_NORM,
+        MODEL_TENSOR.OUTPUT,
+        MODEL_TENSOR.ATTN_NORM,
+        MODEL_TENSOR.ATTN_QKV,
+        MODEL_TENSOR.ATTN_Q,
+        MODEL_TENSOR.ATTN_K,
+        MODEL_TENSOR.ATTN_V,
+        MODEL_TENSOR.ATTN_OUT,
+        MODEL_TENSOR.FFN_NORM,
+        MODEL_TENSOR.FFN_DOWN,
+        MODEL_TENSOR.FFN_UP,
+    ],
     MODEL_ARCH.CODESHELL: [
         MODEL_TENSOR.TOKEN_EMBD,
         MODEL_TENSOR.POS_EMBD,
@@ -837,8 +853,7 @@ class GGUFValueType(IntEnum):
             return GGUFValueType.INT32
         # TODO: need help with 64-bit types in Python
         else:
-            print("Unknown type:", type(val))
-            sys.exit()
+            raise ValueError(f"Unknown type: {type(val)}")
 
 
 # Note: Does not support GGML_QKK_64
@@ -924,6 +939,7 @@ KEY_SSM_TIME_STEP_RANK = Keys.SSM.TIME_STEP_RANK
 
 # tokenization
 KEY_TOKENIZER_MODEL      = Keys.Tokenizer.MODEL
+KEY_TOKENIZER_PRE        = Keys.Tokenizer.PRE
 KEY_TOKENIZER_LIST       = Keys.Tokenizer.LIST
 KEY_TOKENIZER_TOKEN_TYPE = Keys.Tokenizer.TOKEN_TYPE
 KEY_TOKENIZER_SCORES     = Keys.Tokenizer.SCORES

bigdl/cpp/gguf-py/gguf/gguf_reader.py

@@ -4,6 +4,7 @@
 #
 from __future__ import annotations
 
+import logging
 import os
 from collections import OrderedDict
 from typing import Any, Literal, NamedTuple, TypeVar, Union
@@ -27,6 +28,7 @@ from gguf.constants import (
     GGUFValueType,
 )
 
+logger = logging.getLogger(__name__)
 
 READER_SUPPORTED_VERSIONS = [2, GGUF_VERSION]
 
@@ -139,8 +141,13 @@ class GGUFReader:
 
     def _push_field(self, field: ReaderField, skip_sum: bool = False) -> int:
         if field.name in self.fields:
-            raise KeyError(f'Duplicate {field.name} already in list at offset {field.offset}')
-        self.fields[field.name] = field
+            # TODO: add option to generate error on duplicate keys
+            # raise KeyError(f'Duplicate {field.name} already in list at offset {field.offset}')
+
+            logger.warning(f'Duplicate key {field.name} at offset {field.offset}')
+            self.fields[field.name + '_{}'.format(field.offset)] = field
+        else:
+            self.fields[field.name] = field
         return 0 if skip_sum else sum(int(part.nbytes) for part in field.parts)
 
     def _get_str(self, offset: int) -> tuple[npt.NDArray[np.uint64], npt.NDArray[np.uint8]]:
@@ -234,8 +241,14 @@ class GGUFReader:
 
     def _build_tensors(self, start_offs: int, fields: list[ReaderField]) -> None:
         tensors = []
+        tensor_names = set() # keep track of name to prevent duplicated tensors
         for field in fields:
             _name_len, name_data, _n_dims, dims, raw_dtype, offset_tensor = field.parts
+            # check if there's any tensor having same name already in the list
+            tensor_name = str(bytes(name_data), encoding = 'utf-8')
+            if tensor_name in tensor_names:
+                raise ValueError(f'Found duplicated tensor with name {tensor_name}')
+            tensor_names.add(tensor_name)
             ggml_type = GGMLQuantizationType(raw_dtype[0])
             n_elems = np.prod(dims)
             block_size, type_size = GGML_QUANT_SIZES[ggml_type]
@@ -267,7 +280,7 @@ class GGUFReader:
                 item_count = n_bytes
                 item_type = np.uint8
             tensors.append(ReaderTensor(
-                name = str(bytes(name_data), encoding = 'utf-8'),
+                name = tensor_name,
                 tensor_type = ggml_type,
                 shape = dims,
                 n_elements = n_elems,

bigdl/cpp/gguf-py/gguf/gguf_writer.py

@@ -1,5 +1,6 @@
 from __future__ import annotations
 
+import logging
 import os
 import shutil
 import struct
@@ -24,6 +25,8 @@ from .constants import (
     TokenType,
 )
 
+logger = logging.getLogger(__name__)
+
 
 class WriterState(Enum):
     EMPTY   = auto()
@@ -63,10 +66,11 @@ class GGUFWriter:
         self.kv_data_count = 0
         self.ti_data = bytearray()
         self.ti_data_count = 0
+        self.ti_names = set()
         self.use_temp_file = use_temp_file
         self.temp_file = None
         self.tensors = []
-        print("gguf: This GGUF file is for {0} Endian only".format(
+        logger.info("gguf: This GGUF file is for {0} Endian only".format(
             "Big" if self.endianess == GGUFEndian.BIG else "Little",
         ))
         self.state = WriterState.EMPTY
@@ -197,6 +201,10 @@ class GGUFWriter:
         if self.state is not WriterState.EMPTY:
             raise ValueError(f'Expected output file to be empty, got {self.state}')
 
+        if name in self.ti_names:
+            raise ValueError(f'Duplicated tensor name {name}')
+        self.ti_names.add(name)
+
         encoded_name = name.encode("utf8")
         self.ti_data += self._pack("Q", len(encoded_name))
         self.ti_data += encoded_name
@@ -422,6 +430,9 @@ class GGUFWriter:
     def add_tokenizer_model(self, model: str) -> None:
         self.add_string(Keys.Tokenizer.MODEL, model)
 
+    def add_tokenizer_pre(self, pre: str) -> None:
+        self.add_string(Keys.Tokenizer.PRE, pre)
+
     def add_token_list(self, tokens: Sequence[str] | Sequence[bytes] | Sequence[bytearray]) -> None:
         self.add_array(Keys.Tokenizer.LIST, tokens)
 

bigdl/cpp/gguf-py/gguf/tensor_mapping.py

@@ -117,6 +117,7 @@ class TensorNameMap:
             "h.{bid}.attn.c_attn",                                                 # gpt2
             "transformer.h.{bid}.mixer.Wqkv",                                      # phi2
             "encoder.layers.{bid}.attn.Wqkv",                                      # nomic-bert
+            "model.layers.{bid}.self_attn.qkv_proj"                                # phi3
         ),
 
         # Attention query
@@ -234,6 +235,7 @@ class TensorNameMap:
             "h.{bid}.mlp.c_fc",                                       # gpt2
             "transformer.h.{bid}.mlp.fc1",                            # phi2
             "model.layers.{bid}.mlp.fc1",                             # phi2
+            "model.layers.{bid}.mlp.gate_up_proj",                    # phi3
             "model.layers.layers.{bid}.mlp.up_proj",                  # plamo
             "model.layers.{bid}.feed_forward.w3",                     # internlm2
             "encoder.layers.{bid}.mlp.fc11",                          # nomic-bert

bigdl/cpp/gguf-py/gguf/vocab.py

@@ -1,13 +1,15 @@
 from __future__ import annotations
 
+import logging
 import json
 import os
-import sys
 from pathlib import Path
 from typing import Any, Callable
 
 from .gguf_writer import GGUFWriter
 
+logger = logging.getLogger(__name__)
+
 
 class SpecialVocab:
     merges: list[str]
@@ -40,38 +42,29 @@ class SpecialVocab:
     def add_to_gguf(self, gw: GGUFWriter, quiet: bool = False) -> None:
         if self.merges:
             if not quiet:
-                print(f'gguf: Adding {len(self.merges)} merge(s).')
+                logger.info(f'Adding {len(self.merges)} merge(s).')
             gw.add_token_merges(self.merges)
         elif self.load_merges:
-            print(
-                'gguf: WARNING: Adding merges requested but no merges found, output may be non-functional.',
-                file = sys.stderr,
-            )
+            logger.warning('Adding merges requested but no merges found, output may be non-functional.')
         for typ, tokid in self.special_token_ids.items():
             id_handler: Callable[[int], None] | None = getattr(gw, f'add_{typ}_token_id', None)
             if id_handler is None:
-                print(
-                    f'gguf: WARNING: No handler for special token type {typ} with id {tokid} - skipping',
-                    file = sys.stderr,
-                )
+                logger.warning(f'No handler for special token type {typ} with id {tokid} - skipping')
                 continue
             if not quiet:
-                print(f'gguf: Setting special token type {typ} to {tokid}')
+                logger.info(f'Setting special token type {typ} to {tokid}')
             id_handler(tokid)
         for typ, value in self.add_special_token.items():
             add_handler: Callable[[bool], None] | None = getattr(gw, f'add_add_{typ}_token', None)
             if add_handler is None:
-                print(
-                    f'gguf: WARNING: No handler for add_{typ}_token with value {value} - skipping',
-                    file = sys.stderr,
-                )
+                logger.warning(f'No handler for add_{typ}_token with value {value} - skipping')
                 continue
             if not quiet:
-                print(f'gguf: Setting add_{typ}_token to {value}')
+                logger.info(f'Setting add_{typ}_token to {value}')
             add_handler(value)
         if self.chat_template is not None:
             if not quiet:
-                print(f'gguf: Setting chat_template to {self.chat_template}')
+                logger.info(f'Setting chat_template to {self.chat_template}')
             gw.add_chat_template(self.chat_template)
 
     def _load(self, path: Path) -> None:
@@ -99,10 +92,7 @@ class SpecialVocab:
                     continue
                 parts = line.split(None, 3)
                 if len(parts) != 2:
-                    print(
-                        f'gguf: WARNING: {merges_file.name}: Line {line_num}: Entry malformed, ignoring',
-                        file = sys.stderr,
-                    )
+                    logger.warning(f'{merges_file.name}: Line {line_num}: Entry malformed, ignoring')
                     continue
                 merges.append(f'{parts[0]} {parts[1]}')
         self.merges = merges
@@ -118,10 +108,7 @@ class SpecialVocab:
                 return
             self.special_token_ids[typ] = tid
             return
-        print(
-            f'gguf: WARNING: Special token type {typ}, id {tid} out of range, must be under {self.n_vocab} - skipping',
-            file = sys.stderr,
-        )
+        logger.warning(f'Special token type {typ}, id {tid} out of range, must be under {self.n_vocab} - skipping')
 
     def _try_load_from_tokenizer_json(self, path: Path) -> bool:
         tokenizer_file = path / 'tokenizer.json'
@@ -144,10 +131,7 @@ class SpecialVocab:
         if chat_template is None or isinstance(chat_template, (str, list)):
             self.chat_template = chat_template
         else:
-            print(
-                f'gguf: WARNING: Bad type for chat_template field in {tokenizer_config_file!r} - ignoring',
-                file = sys.stderr
-            )
+            logger.warning(f'Bad type for chat_template field in {tokenizer_config_file!r} - ignoring')
         for typ in self.special_token_types:
             add_entry = tokenizer_config.get(f'add_{typ}_token')
             if isinstance(add_entry, bool):

{bigdl_core_cpp-2.5.0b20240508.data → bigdl_core_cpp-2.5.0b20240510.data}/scripts/init-llama-cpp.bat

@@ -15,4 +15,5 @@ for %%f in (*) do (
 popd
 
 copy "%cpp_dir%\convert.py" .
+copy "%cpp_dir%\convert-hf-to-gguf.py" .
 xcopy /E /I "%cpp_dir%\gguf-py\" .\gguf-py

{bigdl_core_cpp-2.5.0b20240508.dist-info → bigdl_core_cpp-2.5.0b20240510.dist-info}/METADATA

@@ -1,6 +1,6 @@
 Metadata-Version: 2.1
 Name: bigdl-core-cpp
-Version: 2.5.0b20240508
+Version: 2.5.0b20240510
 Summary: Large Language Model Develop Toolkit
 Author: BigDL Authors
 License: Apache License, Version 2.0

bigdl_core_cpp-2.5.0b20240510.dist-info/RECORD

@@ -0,0 +1,55 @@
+bigdl/cpp/__init__.py,sha256=47DEQpj8HBSa-_TImW-5JCeuQeRkm5NMpJWZG3hSuFU,0
+bigdl/cpp/convert-hf-to-gguf.py,sha256=67U2K3ix3i1lps7L57-qiTVu5-VvJQbAXcan4aU0KT4,146672
+bigdl/cpp/convert.py,sha256=OPDsmbVairx1s6ftM-G8MhUpPpEwvF48yS8HhOB7_j4,65023
+bigdl/cpp/gguf-py/__init__.py,sha256=47DEQpj8HBSa-_TImW-5JCeuQeRkm5NMpJWZG3hSuFU,0
+bigdl/cpp/gguf-py/gguf/__init__.py,sha256=2eyRnO5rS0WNuzL_N159IRA66ybYkisoSETMSM-tGBQ,135
+bigdl/cpp/gguf-py/gguf/constants.py,sha256=ovAbNrRVlAchqxDpHU3Cey24Kcq6pEmFTmOu_ZZ71ng,32165
+bigdl/cpp/gguf-py/gguf/gguf.py,sha256=QpLc-xU055W2d7CEFvJp2gLIfGO63bdM24ZndZCH6rw,493
+bigdl/cpp/gguf-py/gguf/gguf_reader.py,sha256=e-yzOSzsKjOfaCzXQ8QsJjBIypdqFdI5NO9tCS3VUZk,12081
+bigdl/cpp/gguf-py/gguf/gguf_writer.py,sha256=TM3UeO9d7Y3_uB3GikB4iqZt0uJw7OydijnUEoiFYAA,20496
+bigdl/cpp/gguf-py/gguf/py.typed,sha256=47DEQpj8HBSa-_TImW-5JCeuQeRkm5NMpJWZG3hSuFU,0
+bigdl/cpp/gguf-py/gguf/tensor_mapping.py,sha256=CxeOv5ncOICvskNibM1AFs24QUI6N3n_trU0Dvy8wu8,22290
+bigdl/cpp/gguf-py/gguf/vocab.py,sha256=YiWxOW4ta0R3OI2CWeF3o7RIvJA6O0jQ9fxHnartxvU,6955
+bigdl/cpp/libs/baby-llama.exe,sha256=iy_KBtnrlYCE9ZG8com7kX_mRZ6M-Ok5_AOpD7TnHI0,198144
+bigdl/cpp/libs/batched-bench.exe,sha256=GxIC_lQLErihen1YzTbNEzJsnzPH2bCE6B093NynnnA,204800
+bigdl/cpp/libs/batched.exe,sha256=HZZ_f893O0dxOs8SsdcFNrh8JzdmY2NzPZ0iUx3Dx7Q,207360
+bigdl/cpp/libs/beam-search.exe,sha256=PLqsCA2LL9OFf7CYgB0RKNYKiatb_DrPbvGp94IqY7A,212480
+bigdl/cpp/libs/benchmark.exe,sha256=FjZzyZkpoIAqGwTtvvxsOw00OuHJPCO15HnGH1IL66I,74240
+bigdl/cpp/libs/convert-llama2c-to-ggml.exe,sha256=xyUHdo4wBGakfbokQQbZ2aOE4m2xaHcqqWqJ4uMfcYA,126464
+bigdl/cpp/libs/embedding.exe,sha256=35h8XlVdx_Sin-fmrjrZV6GjnXmlECYJ0N3p9UqNXsI,628224
+bigdl/cpp/libs/export-lora.exe,sha256=K0coBfqmmau9Ay0-vsyR9jnJh2CDVX92Si0pOFB17VQ,90624
+bigdl/cpp/libs/finetune.exe,sha256=MZGzvkaoD9S1oB4IyI-vGnEAVCWlarKb6JSQ1xXYF5M,301056
+bigdl/cpp/libs/ggml_shared.dll,sha256=Gh5VlAGJXlLONMVBunYQOSqX3qZa0kwuSgxEH0kCpe8,4230656
+bigdl/cpp/libs/gguf.exe,sha256=YZjN8irTL2CxXqE-29SLIl5ZVpPnVZkwEAv0dItlhig,3768832
+bigdl/cpp/libs/gritlm.exe,sha256=aSV_zafse0VeXtzgi9rViBnlN6N4LnxUhjKHNCpujwA,623104
+bigdl/cpp/libs/imatrix.exe,sha256=r96d19DBbYnieJt9TofVBH_HODpmCt6U_EzixUZ56rM,660480
+bigdl/cpp/libs/infill.exe,sha256=empguEM4BLHRqz4cUdRYLNTkC7M_SMTTCkD6qVmGHjA,727040
+bigdl/cpp/libs/llama-bench.exe,sha256=rIvug8l2MEZ1yZrakPPSIG3r_j6tUMC_tbKLzae55E0,330752
+bigdl/cpp/libs/llama.dll,sha256=IAWcUOEDvgdWpw_unCOpA4bweOUbYIDWVXptcKe-uE0,5724672
+bigdl/cpp/libs/llava-cli.exe,sha256=pNpQVy7u9ZCZ0D8wTCyCFlKMNIcl4IwFjYLFqTZR4Y8,909312
+bigdl/cpp/libs/llava_shared.dll,sha256=KKfUYGzLGJEStX6d7lGwXCgeV0DlyMS-C3nJD7U9mjg,4500480
+bigdl/cpp/libs/lookahead.exe,sha256=IzAdr4DHj4KBcdbPTdOVfXc1o_H_Y0F5xq7yU0SyAx8,668672
+bigdl/cpp/libs/lookup.exe,sha256=Ql0oes_th424iswCPsY2RxoQAzEAMEyk7h9F91SYJEM,702976
+bigdl/cpp/libs/ls-sycl-device.exe,sha256=bffY6meqsUl9xEdYGBgCItza7cMlBIuonxQDI98jp4I,10240
+bigdl/cpp/libs/main.exe,sha256=jIjAWLN6Ohjcbxfz2qrCgaSAWrGw8yPfGTBeNdkIwKw,758784
+bigdl/cpp/libs/ollama.exe,sha256=GVcI_iTxoK9-hPIzmR2cwXzvVube_zOj6VE9Xk1587c,63376170
+bigdl/cpp/libs/parallel.exe,sha256=RxHF9GmNGLll6kQJwDYOnp_dZJTjV5kq2nJwE47WkuY,686080
+bigdl/cpp/libs/passkey.exe,sha256=732tT1NsGmyJf7IrUcAlN4Z6G7pleRvqW0lb4gPMhQc,214528
+bigdl/cpp/libs/perplexity.exe,sha256=c26C5sbtzdOM_ge_f7gnjh_-oS8RKn2wAe2ve7vZ9Zg,774656
+bigdl/cpp/libs/q8dot.exe,sha256=_iNuI2TBaAiSdmy_Cf4mafFKtcqB0IXm_jX51hYNtm0,53760
+bigdl/cpp/libs/quantize-stats.exe,sha256=u907hbMNNh__8ZauYTQQsGe02pBtA_aMcAjnnAjF3vs,130560
+bigdl/cpp/libs/quantize.exe,sha256=5wzH8RUsK6Xbo-hyf3OUDBGY7hyDBxfAEwAvFKm7zbw,233984
+bigdl/cpp/libs/save-load-state.exe,sha256=-HyCr9BIszkJ4RtKXis-Actz4hWXvbEmu3359SYQPbE,621056
+bigdl/cpp/libs/server.exe,sha256=dXd4HDqHmPalppQ8hHvqpgsYcD8TSAREaPg5Uq6jcws,1682432
+bigdl/cpp/libs/simple.exe,sha256=jvCmqXL1k9K_U34cB34I2nTRI4lLNUASzx_Pdv6HTts,200704
+bigdl/cpp/libs/speculative.exe,sha256=8Lf3UI9d5bQQH-f790ZGGIVkJf77TTARixQNylTu83E,694784
+bigdl/cpp/libs/tokenize.exe,sha256=mj1Sy-6BkKfzK3lZvhu96_SRzHh8UjL2tP5ydi1gSQA,180224
+bigdl/cpp/libs/train-text-from-scratch.exe,sha256=6ok4o-F74u93BHR87Lo98P5KqaNuLtHOpnr7qMtFGCM,281600
+bigdl/cpp/libs/vdot.exe,sha256=8jsWdkMSNw6njNbAPODXjpjGeOC7nRyvD6ENOc_rpiY,56320
+bigdl_core_cpp-2.5.0b20240510.data/scripts/init-llama-cpp.bat,sha256=13AgSYRyzk6hlSz11dPyCMJeSoW9z8PZ9l85E_5GWxs,528
+bigdl_core_cpp-2.5.0b20240510.data/scripts/init-llama-cpp.ps1,sha256=JFOylLxO4MKpllHhdbPuJ1xHi9azxDpzdJns8JtZpkU,501
+bigdl_core_cpp-2.5.0b20240510.data/scripts/init-ollama.bat,sha256=OVX62Q7KH67z6rQrgQUwEc19E3eiAH-8HAIycDmySYc,350
+bigdl_core_cpp-2.5.0b20240510.dist-info/METADATA,sha256=y1KrEZsEyFlrEyUUkBNI_nMrCpSe4aR827YBROTEc4I,668
+bigdl_core_cpp-2.5.0b20240510.dist-info/WHEEL,sha256=at4xwl6JdXdkZHxdo5ixTwJ7ENtVftSy2wqmsdmo_4U,98
+bigdl_core_cpp-2.5.0b20240510.dist-info/top_level.txt,sha256=iGuLfZARD_qANcIMfy0tbbrC3EtCg6BSiH8icc3dLWs,6
+bigdl_core_cpp-2.5.0b20240510.dist-info/RECORD,,