Add a special header per-client call to parser provider data

llama_stack/providers/adapters/inference/together/__init__.py

@@ -4,7 +4,7 @@
 # This source code is licensed under the terms described in the LICENSE file in
 # the root directory of this source tree.
 
-from .config import TogetherImplConfig
+from .config import TogetherImplConfig, TogetherHeaderExtractor
 
 
 async def get_adapter_impl(config: TogetherImplConfig, _deps):

llama_stack/providers/adapters/inference/together/config.py

@@ -4,9 +4,18 @@
 # This source code is licensed under the terms described in the LICENSE file in
 # the root directory of this source tree.
 
-from llama_models.schema_utils import json_schema_type
 from pydantic import BaseModel, Field
 
+from llama_models.schema_utils import json_schema_type
+
+from llama_stack.distribution.request_headers import annotate_header
+
+
+class TogetherHeaderExtractor(BaseModel):
+    api_key: annotate_header(
+        "X-LlamaStack-Together-ApiKey", str, "The API Key for the request"
+    )
+
 
 @json_schema_type
 class TogetherImplConfig(BaseModel):

llama_stack/providers/registry/inference.py

@@ -63,6 +63,7 @@ def available_providers() -> List[ProviderSpec]:
                 ],
                 module="llama_stack.providers.adapters.inference.together",
                 config_class="llama_stack.providers.adapters.inference.together.TogetherImplConfig",
+                header_extractor_class="llama_stack.providers.adapters.inference.together.TogetherHeaderExtractor",
             ),
         ),
     ]

llama_stack/providers/utils/memory/vector_store.py

@@ -16,6 +16,7 @@ import httpx
 import numpy as np
 from numpy.typing import NDArray
 from pypdf import PdfReader
+from termcolor import cprint
 
 from llama_models.llama3.api.datatypes import *  # noqa: F403
 from llama_models.llama3.api.tokenizer import Tokenizer
@@ -160,6 +161,8 @@ class BankWithIndex:
                 self.bank.config.overlap_size_in_tokens
                 or (self.bank.config.chunk_size_in_tokens // 4),
             )
+            if not chunks:
+                continue
             embeddings = model.encode([x.content for x in chunks]).astype(np.float32)
 
             await self.index.add_chunks(chunks, embeddings)