bigdl-core-cpp 2.6.0b20250203__py3-none-win_amd64.whl → 2.6.0b20250204.post0__py3-none-win_amd64.whl

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

@@ -568,6 +568,9 @@ class GGUFWriter:
     def add_base_model_organization(self, source_id: int, organization: str) -> None:
         self.add_string(Keys.General.BASE_MODEL_ORGANIZATION.format(id=source_id), organization)
 
+    def add_base_model_description(self, source_id: int, description: str) -> None:
+        self.add_string(Keys.General.BASE_MODEL_DESCRIPTION.format(id=source_id), description)
+
     def add_base_model_url(self, source_id: int, url: str) -> None:
         self.add_string(Keys.General.BASE_MODEL_URL.format(id=source_id), url)
 
@@ -580,15 +583,42 @@ class GGUFWriter:
     def add_base_model_repo_url(self, source_id: int, repo_url: str) -> None:
         self.add_string(Keys.General.BASE_MODEL_REPO_URL.format(id=source_id), repo_url)
 
+    def add_dataset_count(self, source_count: int) -> None:
+        self.add_uint32(Keys.General.DATASET_COUNT, source_count)
+
+    def add_dataset_name(self, source_id: int, name: str) -> None:
+        self.add_string(Keys.General.DATASET_NAME.format(id=source_id), name)
+
+    def add_dataset_author(self, source_id: int, author: str) -> None:
+        self.add_string(Keys.General.DATASET_AUTHOR.format(id=source_id), author)
+
+    def add_dataset_version(self, source_id: int, version: str) -> None:
+        self.add_string(Keys.General.DATASET_VERSION.format(id=source_id), version)
+
+    def add_dataset_organization(self, source_id: int, organization: str) -> None:
+        self.add_string(Keys.General.DATASET_ORGANIZATION.format(id=source_id), organization)
+
+    def add_dataset_description(self, source_id: int, description: str) -> None:
+        self.add_string(Keys.General.DATASET_DESCRIPTION.format(id=source_id), description)
+
+    def add_dataset_url(self, source_id: int, url: str) -> None:
+        self.add_string(Keys.General.DATASET_URL.format(id=source_id), url)
+
+    def add_dataset_doi(self, source_id: int, doi: str) -> None:
+        self.add_string(Keys.General.DATASET_DOI.format(id=source_id), doi)
+
+    def add_dataset_uuid(self, source_id: int, uuid: str) -> None:
+        self.add_string(Keys.General.DATASET_UUID.format(id=source_id), uuid)
+
+    def add_dataset_repo_url(self, source_id: int, repo_url: str) -> None:
+        self.add_string(Keys.General.DATASET_REPO_URL.format(id=source_id), repo_url)
+
     def add_tags(self, tags: Sequence[str]) -> None:
         self.add_array(Keys.General.TAGS, tags)
 
     def add_languages(self, languages: Sequence[str]) -> None:
         self.add_array(Keys.General.LANGUAGES, languages)
 
-    def add_datasets(self, datasets: Sequence[str]) -> None:
-        self.add_array(Keys.General.DATASETS, datasets)
-
     def add_tensor_data_layout(self, layout: str) -> None:
         self.add_string(Keys.LLM.TENSOR_DATA_LAYOUT.format(arch=self.arch), layout)
 
@@ -721,6 +751,9 @@ class GGUFWriter:
     def add_rope_dimension_count(self, count: int) -> None:
         self.add_uint32(Keys.Rope.DIMENSION_COUNT.format(arch=self.arch), count)
 
+    def add_rope_dimension_sections(self, dims: Sequence[int]) -> None:
+        self.add_array(Keys.Rope.DIMENSION_SECTIONS.format(arch=self.arch), dims)
+
     def add_rope_freq_base(self, value: float) -> None:
         self.add_float32(Keys.Rope.FREQ_BASE.format(arch=self.arch), value)
 
@@ -843,15 +876,6 @@ class GGUFWriter:
 
         self.add_string(Keys.Tokenizer.CHAT_TEMPLATE, value)
 
-    def add_prefix_token_id(self, id: int) -> None:
-        self.add_uint32(Keys.Tokenizer.PREFIX_ID, id)
-
-    def add_suffix_token_id(self, id: int) -> None:
-        self.add_uint32(Keys.Tokenizer.SUFFIX_ID, id)
-
-    def add_middle_token_id(self, id: int) -> None:
-        self.add_uint32(Keys.Tokenizer.MIDDLE_ID, id)
-
     def add_eot_token_id(self, id: int) -> None:
         self.add_uint32(Keys.Tokenizer.EOT_ID, id)
 

bigdl/cpp/gguf-py/gguf/metadata.py

@@ -41,7 +41,7 @@ class Metadata:
     base_models: Optional[list[dict]] = None
     tags: Optional[list[str]] = None
     languages: Optional[list[str]] = None
-    datasets: Optional[list[str]] = None
+    datasets: Optional[list[dict]] = None
 
     @staticmethod
     def load(metadata_override_path: Optional[Path] = None, model_path: Optional[Path] = None, model_name: Optional[str] = None, total_params: int = 0) -> Metadata:
@@ -91,9 +91,11 @@ class Metadata:
         # Base Models is received here as an array of models
         metadata.base_models     = metadata_override.get("general.base_models",        metadata.base_models)
 
+        # Datasets is received here as an array of datasets
+        metadata.datasets        = metadata_override.get("general.datasets",           metadata.datasets)
+
         metadata.tags            = metadata_override.get(Keys.General.TAGS,            metadata.tags)
         metadata.languages       = metadata_override.get(Keys.General.LANGUAGES,       metadata.languages)
-        metadata.datasets        = metadata_override.get(Keys.General.DATASETS,        metadata.datasets)
 
         # Direct Metadata Override (via direct cli argument)
         if model_name is not None:
@@ -346,12 +348,12 @@ class Metadata:
             use_model_card_metadata("author", "model_creator")
             use_model_card_metadata("basename", "model_type")
 
-            if "base_model" in model_card:
+            if "base_model" in model_card or "base_models" in model_card or "base_model_sources" in model_card:
                 # This represents the parent models that this is based on
                 # Example: stabilityai/stable-diffusion-xl-base-1.0. Can also be a list (for merges)
                 # Example of merges: https://huggingface.co/EmbeddedLLM/Mistral-7B-Merge-14-v0.1/blob/main/README.md
                 metadata_base_models = []
-                base_model_value = model_card.get("base_model", None)
+                base_model_value = model_card.get("base_model", model_card.get("base_models", model_card.get("base_model_sources", None)))
 
                 if base_model_value is not None:
                     if isinstance(base_model_value, str):
@@ -364,18 +366,106 @@ class Metadata:
 
                 for model_id in metadata_base_models:
                     # NOTE: model size of base model is assumed to be similar to the size of the current model
-                    model_full_name_component, org_component, basename, finetune, version, size_label = Metadata.get_model_id_components(model_id, total_params)
                     base_model = {}
-                    if model_full_name_component is not None:
-                        base_model["name"] = Metadata.id_to_title(model_full_name_component)
-                    if org_component is not None:
-                        base_model["organization"] = Metadata.id_to_title(org_component)
-                    if version is not None:
-                        base_model["version"] = version
-                    if org_component is not None and model_full_name_component is not None:
-                        base_model["repo_url"] = f"https://huggingface.co/{org_component}/{model_full_name_component}"
+                    if isinstance(model_id, str):
+                        if model_id.startswith("http://") or model_id.startswith("https://") or model_id.startswith("ssh://"):
+                            base_model["repo_url"] = model_id
+
+                            # Check if Hugging Face ID is present in URL
+                            if "huggingface.co" in model_id:
+                                match = re.match(r"https?://huggingface.co/([^/]+/[^/]+)$", model_id)
+                                if match:
+                                    model_id_component = match.group(1)
+                                    model_full_name_component, org_component, basename, finetune, version, size_label = Metadata.get_model_id_components(model_id_component, total_params)
+
+                                    # Populate model dictionary with extracted components
+                                    if model_full_name_component is not None:
+                                        base_model["name"] = Metadata.id_to_title(model_full_name_component)
+                                    if org_component is not None:
+                                        base_model["organization"] = Metadata.id_to_title(org_component)
+                                    if version is not None:
+                                        base_model["version"] = version
+
+                        else:
+                            # Likely a Hugging Face ID
+                            model_full_name_component, org_component, basename, finetune, version, size_label = Metadata.get_model_id_components(model_id, total_params)
+
+                            # Populate model dictionary with extracted components
+                            if model_full_name_component is not None:
+                                base_model["name"] = Metadata.id_to_title(model_full_name_component)
+                            if org_component is not None:
+                                base_model["organization"] = Metadata.id_to_title(org_component)
+                            if version is not None:
+                                base_model["version"] = version
+                            if org_component is not None and model_full_name_component is not None:
+                                base_model["repo_url"] = f"https://huggingface.co/{org_component}/{model_full_name_component}"
+
+                    elif isinstance(model_id, dict):
+                        base_model = model_id
+
+                    else:
+                        logger.error(f"base model entry '{str(model_id)}' not in a known format")
+
                     metadata.base_models.append(base_model)
 
+            if "datasets" in model_card or "dataset" in model_card or "dataset_sources" in model_card:
+                # This represents the datasets that this was trained from
+                metadata_datasets = []
+                dataset_value = model_card.get("datasets", model_card.get("dataset", model_card.get("dataset_sources", None)))
+
+                if dataset_value is not None:
+                    if isinstance(dataset_value, str):
+                        metadata_datasets.append(dataset_value)
+                    elif isinstance(dataset_value, list):
+                        metadata_datasets.extend(dataset_value)
+
+                if metadata.datasets is None:
+                    metadata.datasets = []
+
+                for dataset_id in metadata_datasets:
+                    # NOTE: model size of base model is assumed to be similar to the size of the current model
+                    dataset = {}
+                    if isinstance(dataset_id, str):
+                        if dataset_id.startswith(("http://", "https://", "ssh://")):
+                            dataset["repo_url"] = dataset_id
+
+                            # Check if Hugging Face ID is present in URL
+                            if "huggingface.co" in dataset_id:
+                                match = re.match(r"https?://huggingface.co/([^/]+/[^/]+)$", dataset_id)
+                                if match:
+                                    dataset_id_component = match.group(1)
+                                    dataset_name_component, org_component, basename, finetune, version, size_label = Metadata.get_model_id_components(dataset_id_component, total_params)
+
+                                    # Populate dataset dictionary with extracted components
+                                    if dataset_name_component is not None:
+                                        dataset["name"] = Metadata.id_to_title(dataset_name_component)
+                                    if org_component is not None:
+                                        dataset["organization"] = Metadata.id_to_title(org_component)
+                                    if version is not None:
+                                        dataset["version"] = version
+
+                        else:
+                            # Likely a Hugging Face ID
+                            dataset_name_component, org_component, basename, finetune, version, size_label = Metadata.get_model_id_components(dataset_id, total_params)
+
+                            # Populate dataset dictionary with extracted components
+                            if dataset_name_component is not None:
+                                dataset["name"] = Metadata.id_to_title(dataset_name_component)
+                            if org_component is not None:
+                                dataset["organization"] = Metadata.id_to_title(org_component)
+                            if version is not None:
+                                dataset["version"] = version
+                            if org_component is not None and dataset_name_component is not None:
+                                dataset["repo_url"] = f"https://huggingface.co/{org_component}/{dataset_name_component}"
+
+                    elif isinstance(dataset_id, dict):
+                        dataset = dataset_id
+
+                    else:
+                        logger.error(f"dataset entry '{str(dataset_id)}' not in a known format")
+
+                    metadata.datasets.append(dataset)
+
             use_model_card_metadata("license", "license")
             use_model_card_metadata("license_name", "license_name")
             use_model_card_metadata("license_link", "license_link")
@@ -386,9 +476,6 @@ class Metadata:
             use_array_model_card_metadata("languages", "languages")
             use_array_model_card_metadata("languages", "language")
 
-            use_array_model_card_metadata("datasets", "datasets")
-            use_array_model_card_metadata("datasets", "dataset")
-
         # Hugging Face Parameter Heuristics
         ####################################
 
@@ -458,7 +545,10 @@ class Metadata:
             gguf_writer.add_size_label(self.size_label)
 
         if self.license is not None:
-            gguf_writer.add_license(self.license)
+            if isinstance(self.license, list):
+                gguf_writer.add_license(",".join(self.license))
+            else:
+                gguf_writer.add_license(self.license)
         if self.license_name is not None:
             gguf_writer.add_license_name(self.license_name)
         if self.license_link is not None:
@@ -493,6 +583,8 @@ class Metadata:
                     gguf_writer.add_base_model_version(key, base_model_entry["version"])
                 if "organization" in base_model_entry:
                     gguf_writer.add_base_model_organization(key, base_model_entry["organization"])
+                if "description" in base_model_entry:
+                    gguf_writer.add_base_model_description(key, base_model_entry["description"])
                 if "url" in base_model_entry:
                     gguf_writer.add_base_model_url(key, base_model_entry["url"])
                 if "doi" in base_model_entry:
@@ -502,9 +594,29 @@ class Metadata:
                 if "repo_url" in base_model_entry:
                     gguf_writer.add_base_model_repo_url(key, base_model_entry["repo_url"])
 
+        if self.datasets is not None:
+            gguf_writer.add_dataset_count(len(self.datasets))
+            for key, dataset_entry in enumerate(self.datasets):
+                if "name" in dataset_entry:
+                    gguf_writer.add_dataset_name(key, dataset_entry["name"])
+                if "author" in dataset_entry:
+                    gguf_writer.add_dataset_author(key, dataset_entry["author"])
+                if "version" in dataset_entry:
+                    gguf_writer.add_dataset_version(key, dataset_entry["version"])
+                if "organization" in dataset_entry:
+                    gguf_writer.add_dataset_organization(key, dataset_entry["organization"])
+                if "description" in dataset_entry:
+                    gguf_writer.add_dataset_description(key, dataset_entry["description"])
+                if "url" in dataset_entry:
+                    gguf_writer.add_dataset_url(key, dataset_entry["url"])
+                if "doi" in dataset_entry:
+                    gguf_writer.add_dataset_doi(key, dataset_entry["doi"])
+                if "uuid" in dataset_entry:
+                    gguf_writer.add_dataset_uuid(key, dataset_entry["uuid"])
+                if "repo_url" in dataset_entry:
+                    gguf_writer.add_dataset_repo_url(key, dataset_entry["repo_url"])
+
         if self.tags is not None:
             gguf_writer.add_tags(self.tags)
         if self.languages is not None:
             gguf_writer.add_languages(self.languages)
-        if self.datasets is not None:
-            gguf_writer.add_datasets(self.datasets)

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

@@ -13,7 +13,7 @@ class TensorNameMap:
             "transformer.wte",                           # gpt2 gpt-j mpt refact qwen dbrx jais exaone
             "transformer.word_embeddings",               # falcon
             "word_embeddings",                           # bloom
-            "model.embed_tokens",                        # llama-hf nemotron olmoe
+            "model.embed_tokens",                        # llama-hf nemotron olmoe olmo2
             "tok_embeddings",                            # llama-pth
             "embeddings.word_embeddings",                # bert nomic-bert
             "language_model.embedding.word_embeddings",  # persimmon
@@ -54,7 +54,7 @@ class TensorNameMap:
         # Output
         MODEL_TENSOR.OUTPUT: (
             "embed_out",                 # gptneox
-            "lm_head",                   # gpt2 mpt falcon llama-hf baichuan qwen mamba dbrx jais nemotron exaone olmoe
+            "lm_head",                   # gpt2 mpt falcon llama-hf baichuan qwen mamba dbrx jais nemotron exaone olmoe olmo2
             "output",                    # llama-pth bloom internlm2
             "word_embeddings_for_head",  # persimmon
             "lm_head.linear",            # phi2
@@ -66,7 +66,7 @@ class TensorNameMap:
         MODEL_TENSOR.OUTPUT_NORM: (
             "gpt_neox.final_layer_norm",               # gptneox
             "transformer.ln_f",                        # gpt2 gpt-j falcon jais exaone
-            "model.norm",                              # llama-hf baichuan internlm2 olmoe
+            "model.norm",                              # llama-hf baichuan internlm2 olmoe olmo2
             "norm",                                    # llama-pth
             "transformer.norm_f",                      # mpt dbrx
             "ln_f",                                    # refact bloom qwen gpt2
@@ -145,7 +145,8 @@ class TensorNameMap:
 
         # Attention query
         MODEL_TENSOR.ATTN_Q: (
-            "model.layers.{bid}.self_attn.q_proj",                       # llama-hf nemotron olmoe
+            "model.layers.{bid}.self_attn.q_proj",                       # llama-hf nemotron olmoe olmo2
+            "model.layers.{bid}.self_attn.q_proj_no_perm",               # llama-custom
             "layers.{bid}.attention.wq",                                 # llama-pth
             "encoder.layer.{bid}.attention.self.query",                  # bert
             "transformer.h.{bid}.attn.q_proj",                           # gpt-j
@@ -157,7 +158,8 @@ class TensorNameMap:
 
         # Attention key
         MODEL_TENSOR.ATTN_K: (
-            "model.layers.{bid}.self_attn.k_proj",                     # llama-hf nemotron olmoe
+            "model.layers.{bid}.self_attn.k_proj",                     # llama-hf nemotron olmoe olmo2
+            "model.layers.{bid}.self_attn.k_proj_no_perm",             # llama-custom
             "layers.{bid}.attention.wk",                               # llama-pth
             "encoder.layer.{bid}.attention.self.key",                  # bert
             "transformer.h.{bid}.attn.k_proj",                         # gpt-j
@@ -170,7 +172,7 @@ class TensorNameMap:
 
         # Attention value
         MODEL_TENSOR.ATTN_V: (
-            "model.layers.{bid}.self_attn.v_proj",                       # llama-hf nemotron olmoe
+            "model.layers.{bid}.self_attn.v_proj",                       # llama-hf nemotron olmoe olmo2
             "layers.{bid}.attention.wv",                                 # llama-pth
             "encoder.layer.{bid}.attention.self.value",                  # bert
             "transformer.h.{bid}.attn.v_proj",                           # gpt-j
@@ -188,7 +190,7 @@ class TensorNameMap:
             "transformer.blocks.{bid}.attn.out_proj",                       # mpt
             "transformer.h.{bid}.self_attention.dense",                     # falcon
             "h.{bid}.self_attention.dense",                                 # bloom
-            "model.layers.{bid}.self_attn.o_proj",                          # llama-hf nemotron olmoe
+            "model.layers.{bid}.self_attn.o_proj",                          # llama-hf nemotron olmoe olmo2
             "layers.{bid}.attention.wo",                                    # llama-pth
             "encoder.layer.{bid}.attention.output.dense",                   # bert
             "transformer.h.{bid}.attn.out_proj",                            # gpt-j
@@ -215,7 +217,7 @@ class TensorNameMap:
         ),
 
         MODEL_TENSOR.ATTN_POST_NORM: (
-            "model.layers.{bid}.post_attention_layernorm",     # gemma2
+            "model.layers.{bid}.post_attention_layernorm",     # gemma2 olmo2
         ),
 
         # Rotary embeddings
@@ -250,7 +252,7 @@ class TensorNameMap:
 
         # Post feed-forward norm
         MODEL_TENSOR.FFN_POST_NORM: (
-            "model.layers.{bid}.post_feedforward_layernorm", # gemma2
+            "model.layers.{bid}.post_feedforward_layernorm", # gemma2 olmo2
         ),
 
         MODEL_TENSOR.FFN_GATE_INP: (
@@ -273,7 +275,7 @@ class TensorNameMap:
             "transformer.blocks.{bid}.ffn.up_proj",                   # mpt
             "transformer.h.{bid}.mlp.dense_h_to_4h",                  # falcon
             "h.{bid}.mlp.dense_h_to_4h",                              # bloom
-            "model.layers.{bid}.mlp.up_proj",                         # llama-hf refact nemotron
+            "model.layers.{bid}.mlp.up_proj",                         # llama-hf refact nemotron olmo2
             "layers.{bid}.feed_forward.w3",                           # llama-pth
             "encoder.layer.{bid}.intermediate.dense",                 # bert
             "transformer.h.{bid}.mlp.fc_in",                          # gpt-j
@@ -314,7 +316,7 @@ class TensorNameMap:
 
         # Feed-forward gate
         MODEL_TENSOR.FFN_GATE: (
-            "model.layers.{bid}.mlp.gate_proj",           # llama-hf refact
+            "model.layers.{bid}.mlp.gate_proj",           # llama-hf refact olmo2
             "layers.{bid}.feed_forward.w1",               # llama-pth
             "transformer.h.{bid}.mlp.w2",                 # qwen
             "transformer.h.{bid}.mlp.c_fc2",              # jais
@@ -346,7 +348,7 @@ class TensorNameMap:
             "transformer.blocks.{bid}.ffn.down_proj",                 # mpt
             "transformer.h.{bid}.mlp.dense_4h_to_h",                  # falcon
             "h.{bid}.mlp.dense_4h_to_h",                              # bloom
-            "model.layers.{bid}.mlp.down_proj",                       # llama-hf nemotron
+            "model.layers.{bid}.mlp.down_proj",                       # llama-hf nemotron olmo2
             "layers.{bid}.feed_forward.w2",                           # llama-pth
             "encoder.layer.{bid}.output.dense",                       # bert
             "transformer.h.{bid}.mlp.fc_out",                         # gpt-j
@@ -383,7 +385,7 @@ class TensorNameMap:
         MODEL_TENSOR.ATTN_Q_NORM: (
             "language_model.encoder.layers.{bid}.self_attention.q_layernorm",
             "model.layers.{bid}.self_attn.q_layernorm",                       # persimmon
-            "model.layers.{bid}.self_attn.q_norm",                            # cohere olmoe chameleon
+            "model.layers.{bid}.self_attn.q_norm",                            # cohere olmoe chameleon olmo2
             "transformer.blocks.{bid}.attn.q_ln",                             # sea-lion
             "encoder.layer.{bid}.attention.self.layer_norm_q",                # jina-bert-v2
             "transformer.layers.{bid}.attn.q_norm",                           # openelm
@@ -392,7 +394,7 @@ class TensorNameMap:
         MODEL_TENSOR.ATTN_K_NORM: (
             "language_model.encoder.layers.{bid}.self_attention.k_layernorm",
             "model.layers.{bid}.self_attn.k_layernorm",                       # persimmon
-            "model.layers.{bid}.self_attn.k_norm",                            # cohere olmoe chameleon
+            "model.layers.{bid}.self_attn.k_norm",                            # cohere olmoe chameleon olmo2
             "transformer.blocks.{bid}.attn.k_ln",                             # sea-lion
             "encoder.layer.{bid}.attention.self.layer_norm_k",                # jina-bert-v2
             "transformer.layers.{bid}.attn.k_norm",                           # openelm
@@ -766,4 +768,4 @@ class TensorNameMap:
 
 
 def get_tensor_name_map(arch: MODEL_ARCH, n_blocks: int) -> TensorNameMap:
-    return TensorNameMap(arch, n_blocks)
+    return TensorNameMap(arch, n_blocks)

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

@@ -122,8 +122,30 @@ class SpecialVocab:
                 tokenizer = json.load(f)
             if self.load_merges:
                 merges = tokenizer.get('model', {}).get('merges')
-                if isinstance(merges, list) and merges and isinstance(merges[0], str):
-                    self.merges = merges
+                if isinstance(merges, list) and merges:
+                    if isinstance(merges[0], str):
+                        self.merges = merges
+                    elif isinstance(merges[0], list) and len(merges[0]) == 2 and isinstance(merges[0][0], str):
+                        # New format since transformers 4.45 to support spaces in merges
+                        # ref: https://github.com/ggerganov/llama.cpp/issues/9692
+                        # TODO: internally store as the new format instead of converting to old
+                        if any(' ' in s for pair in merges for s in pair):
+                            logger.warning(f'Spaces in merges detected, encoding as {chr(ord(" ") + 256)!r}')
+                        self.merges = [
+                            ' '.join(
+                                [
+                                    # ensure the spaces are properly encoded
+                                    ''.join(
+                                        chr(ord(c) + 256) if c == ' ' else c
+                                        for c in part
+                                    )
+                                    for part in pair
+                                ]
+                            )
+                            for pair in merges
+                        ]
+                    else:
+                        raise ValueError("Unknown tokenizer merges format")
             added_tokens = tokenizer.get('added_tokens', {})
         else:
             added_tokens = {}

{bigdl_core_cpp-2.6.0b20250203.data → bigdl_core_cpp-2.6.0b20250204.post0.data}/scripts/init-ollama.bat

@@ -6,7 +6,7 @@ set "cpp_dir=%cpp_dir:~0,-1%"
 set "lib_dir=%cpp_dir%\libs"
 
 :: Create symlinks for DLLs and EXE
-for %%f in (ollama.exe ollama_llama.dll ollama_ggml.dll ollama_llava_shared.dll libc++.dll) do (
+for %%f in (ollama.exe ollama_llama.dll ollama_ggml.dll ollama_llava_shared.dll ollama-ggml-base.dll ollama-ggml-cpu.dll ollama-ggml-sycl.dll libc++.dll) do (
     if exist "%cd%\%%f" del /f "%cd%\%%f"
     mklink "%cd%\%%f" "%lib_dir%\%%f"
 )

{bigdl_core_cpp-2.6.0b20250203.dist-info → bigdl_core_cpp-2.6.0b20250204.post0.dist-info}/METADATA

@@ -1,6 +1,6 @@
 Metadata-Version: 2.2
 Name: bigdl-core-cpp
-Version: 2.6.0b20250203
+Version: 2.6.0b20250204.post0
 Summary: Large Language Model Develop Toolkit
 Author: BigDL Authors
 License: Apache License, Version 2.0

bigdl_core_cpp-2.6.0b20250204.post0.dist-info/RECORD

@@ -0,0 +1,54 @@
+bigdl/cpp/__init__.py,sha256=47DEQpj8HBSa-_TImW-5JCeuQeRkm5NMpJWZG3hSuFU,0
+bigdl/cpp/convert_hf_to_gguf.py,sha256=HVJ6axht-K3DAmmkVkC1T1_vZJ8FprUwfNpXJKnLsUQ,210336
+bigdl/cpp/convert_hf_to_gguf_update.py,sha256=4A9Q4oLh5tZAovmgKgS7bVlcCqcVm4j0SIwGK_lww9s,17004
+bigdl/cpp/convert_llama_ggml_to_gguf.py,sha256=0dKjRhmFzvWV4e-cuLmaeW14JrWUtZwerBmz8mYyMvI,19556
+bigdl/cpp/convert_lora_to_gguf.py,sha256=b2CUmTK-ztrJE_50DzsXK3SRZshr_LYFyUPn0UDnkiA,17270
+bigdl/cpp/gguf-py/__init__.py,sha256=47DEQpj8HBSa-_TImW-5JCeuQeRkm5NMpJWZG3hSuFU,0
+bigdl/cpp/gguf-py/gguf/__init__.py,sha256=h5GWs6SMXYR8giWZ7MTZzAc3hYsIJF-HAkdxtgXLOPo,228
+bigdl/cpp/gguf-py/gguf/constants.py,sha256=hN1QPaL74Ef8iJXFE2hlRJmG1w43a2E6HwB0xCxp8ic,61425
+bigdl/cpp/gguf-py/gguf/gguf.py,sha256=QpLc-xU055W2d7CEFvJp2gLIfGO63bdM24ZndZCH6rw,493
+bigdl/cpp/gguf-py/gguf/gguf_reader.py,sha256=PUrx08ZwaUOz1gLw5JQ459Hi7JIeCdlHgZX7wXcTqbI,12702
+bigdl/cpp/gguf-py/gguf/gguf_writer.py,sha256=e-8gwsdq0sipd8zzrXvvtVWV7mCaQD9fRsCn6_67CNs,38541
+bigdl/cpp/gguf-py/gguf/lazy.py,sha256=YIYxGBWD-oKXU4HOvpHs9eiEn81HUgeSmt1mmHJlbdM,8814
+bigdl/cpp/gguf-py/gguf/metadata.py,sha256=oBTb4DXi_h1L_gYm8x_JRVuEPR4GHlVHuM-iN0OxWoY,33244
+bigdl/cpp/gguf-py/gguf/py.typed,sha256=47DEQpj8HBSa-_TImW-5JCeuQeRkm5NMpJWZG3hSuFU,0
+bigdl/cpp/gguf-py/gguf/quants.py,sha256=2z6vcK-kBefqZbYNmSEVmdZF_tXHeVb5NC6jCbBdgKc,62040
+bigdl/cpp/gguf-py/gguf/tensor_mapping.py,sha256=tUkpDyeMXwf9M1Cx1HL2g6mMavngsj0UYkeWMaOPjQU,35561
+bigdl/cpp/gguf-py/gguf/utility.py,sha256=LAwLstUlosYQ1oX9LlJZ-1uCmwyNtOFcJfXpkLnju0k,3003
+bigdl/cpp/gguf-py/gguf/vocab.py,sha256=ulUC8XudFDuZC2SNKGNQpsU2KFP_YhkyNLIWV16jG6I,20816
+bigdl/cpp/libs/common.lib,sha256=ELuS4xoqHJYsjeta8zWKSOKDZHqrnAwuWa3blD7O4F8,3955904
+bigdl/cpp/libs/ggml-cpu.dll,sha256=WxBkX7Smps1FOcCaDfzxMAKlUgLTqfKvVb7kEt2GAdo,475648
+bigdl/cpp/libs/ggml-sycl.dll,sha256=y0xMUN5r3kn1idTnWep_HOSp3OEMByiq-MYfFUl5Fms,5280768
+bigdl/cpp/libs/ggml.dll,sha256=A8buG4Wm9lJ4byLJAaqfrjacjigfq5aLuoOMU3Gs0Lw,113152
+bigdl/cpp/libs/libc++.dll,sha256=U0TVK2WfFQIJPP6Bz9SeJmgskm2iqZWJorx_DGdfKIw,1561600
+bigdl/cpp/libs/llama-batched.exe,sha256=OXRvTB7SgBF3kSzluFaZ9T6AYLcRIJ-3fDGBQYMSADM,847872
+bigdl/cpp/libs/llama-bench.exe,sha256=MZM1q--5V9DMr_rQjRpe54Hjbj1hzh6B87QqA_cHS0s,277504
+bigdl/cpp/libs/llama-cli.exe,sha256=SbG_NSl6VY7ChmwRD2RYgxO-QI8tiTfzeoj3tq8WkHE,922624
+bigdl/cpp/libs/llama-embedding.exe,sha256=msgusVBW43uLEBxq26Vjg8KKoiHPBGNiCvQR4s7HnfE,870400
+bigdl/cpp/libs/llama-gguf.exe,sha256=ijuXVYdZSeRcDr35M-DmKS5sJQ0Ab_n_b0z5SQU_x9k,58880
+bigdl/cpp/libs/llama-llava-cli.exe,sha256=fqgtEg_c2uXUTgmGQWJ9pFP0ie75_GkswP56qBWGZ4w,1109504
+bigdl/cpp/libs/llama-lookup.exe,sha256=QCPdyiI8AsFFXjAnX1GohxOiFKqPRLoL7_G9K2ihN7Y,904704
+bigdl/cpp/libs/llama-ls-sycl-device.exe,sha256=IfpDycbUBZgmohCtMUNI3k5hDpg1QkqimyVuYPMk_dw,10240
+bigdl/cpp/libs/llama-minicpmv-cli.exe,sha256=j-Tk3T7Kb6N0ccR2F-UTBPuxjFYHfSs5WbfSc9HAGhQ,1107456
+bigdl/cpp/libs/llama-perplexity.exe,sha256=ZEEjyN3JnTjLBwb2DtM4SOHFbgH-HzdfJlKZX_sQIhw,990720
+bigdl/cpp/libs/llama-quantize.exe,sha256=5bKWjuDw3kbjCuwN0iPOobph6rfU-j0os_dYOFG0zHI,122880
+bigdl/cpp/libs/llama-server.exe,sha256=32cg0YZkZTJQ24Vst7kO3RhCDN1BoOrUQUwBVPJgUes,2148864
+bigdl/cpp/libs/llama-simple.exe,sha256=nIQ7VJ_I1Fqc9gsCDt_BVx8M-lIiLh_byycNPzH6lqw,62464
+bigdl/cpp/libs/llama-speculative.exe,sha256=xeDoKmoJPae-r1FBTApS-z5K1AhWA13lbpUQdqGjSpg,907264
+bigdl/cpp/libs/llama-tokenize.exe,sha256=cJLtVEuYfli42CJYGet0beV6m6pW7r2m44QtgOqlRt0,88576
+bigdl/cpp/libs/llama.dll,sha256=2Wgwo2UBczq2s6SBP9TBZMxJMmr_P8CfobB0NTlSkis,1404928
+bigdl/cpp/libs/llava_shared.dll,sha256=vBVJuzZbXjf0W14a5ZO415Wj1-qiwT_Qcex-v71RjVQ,365056
+bigdl/cpp/libs/ollama-ggml-base.dll,sha256=2yGQxfoge3KDesOWcwa8ncFMvyvg7Jd8pl_pvMN6w5w,459776
+bigdl/cpp/libs/ollama-ggml-cpu.dll,sha256=iQEVXQK_tHmGY4UpMrPPj6G9ssbPxHeJQR0s5rqG6cU,475648
+bigdl/cpp/libs/ollama-ggml-sycl.dll,sha256=Y_0PbkDf48ypCVA71QLXap_Wgbd5n76m4gKQEt64NFk,5280768
+bigdl/cpp/libs/ollama.exe,sha256=tTdj_U3k2XHx4R2e7Z_tTOaqnOtQAJQqWNdo6oxTnHU,25979904
+bigdl/cpp/libs/ollama_ggml.dll,sha256=TPiPOCTK2Lfcf4X03NIVwiYTMPeYAKX9VTmeI9b3RKc,113152
+bigdl/cpp/libs/ollama_llama.dll,sha256=UGD_a9ok5m0vnLGtF2cSHwnQJRAV6XaGDBMMdIT0g00,1427968
+bigdl/cpp/libs/ollama_llava_shared.dll,sha256=3e7S_cVWEvfUUEUYWUmTpeDsMS7Wr5_Hofc4lAUmK_w,365056
+bigdl_core_cpp-2.6.0b20250204.post0.data/scripts/init-llama-cpp.bat,sha256=U0h6RifZxL3GGJp-0dxdZapQIvXUATSj644CURJL-lg,751
+bigdl_core_cpp-2.6.0b20250204.post0.data/scripts/init-llama-cpp.ps1,sha256=JFOylLxO4MKpllHhdbPuJ1xHi9azxDpzdJns8JtZpkU,501
+bigdl_core_cpp-2.6.0b20250204.post0.data/scripts/init-ollama.bat,sha256=DKWd9wFMT7hcNVQViMxG3CnqF4iZXCh2BmuS3zQjAuw,642
+bigdl_core_cpp-2.6.0b20250204.post0.dist-info/METADATA,sha256=mLLb2aUYHIfuJCKqAxZ2Wy1s8hPohkWVoh8m2gGrn90,756
+bigdl_core_cpp-2.6.0b20250204.post0.dist-info/WHEEL,sha256=2wr--P33L_Xt79Mrb57-zn6CrTlNaEVHEwbOduMxJRg,97
+bigdl_core_cpp-2.6.0b20250204.post0.dist-info/top_level.txt,sha256=iGuLfZARD_qANcIMfy0tbbrC3EtCg6BSiH8icc3dLWs,6
+bigdl_core_cpp-2.6.0b20250204.post0.dist-info/RECORD,,