neuralnode 2.2.1__tar.gz → 2.2.2__tar.gz

{neuralnode-2.2.1 → neuralnode-2.2.2}/PKG-INFO

@@ -1,6 +1,6 @@
 Metadata-Version: 2.4
 Name: neuralnode
-Version: 2.2.1
+Version: 2.2.2
 Summary: Comprehensive AI Framework with 50+ LLM Providers, Advanced Agents, Chains, Memory, RAG, and 100+ Tools
 Project-URL: Homepage, https://assem.cloud/
 Project-URL: Documentation, https://neuralnode.readthedocs.io

{neuralnode-2.2.1 → neuralnode-2.2.2}/pyproject.toml

@@ -4,7 +4,7 @@ build-backend = "hatchling.build"
 
 [project]
 name = "neuralnode"
-version = "2.2.1"
+version = "2.2.2"
 description = "Comprehensive AI Framework with 50+ LLM Providers, Advanced Agents, Chains, Memory, RAG, and 100+ Tools"
 readme = "README.md"
 requires-python = ">=3.9"

{neuralnode-2.2.1 → neuralnode-2.2.2}/src/neuralnode/__init__.py

@@ -42,7 +42,7 @@ Quick Start::
     text = sr.listen()
 """
 
-__version__ = "2.2.1"
+__version__ = "2.2.2"
 __author__ = "NeuralNode Contributors"
 
 # ── Core types ────────────────────────────────────────────────────────────────

{neuralnode-2.2.1 → neuralnode-2.2.2}/src/neuralnode/providers/horus.py

@@ -438,6 +438,7 @@ class HorusProvider(BaseLLMProvider):
         model_id: str = "horus",
         device: Optional[str] = None,
         n_gpu_layers: Optional[int] = None,
+        n_ctx: Optional[int] = None,
         torch_dtype: Optional[Any] = None,
         load_in_8bit: bool = False,
         load_in_4bit: bool = False,
@@ -518,6 +519,7 @@ class HorusProvider(BaseLLMProvider):
         self.model_id = model_id
         self.device = device or self._resolve_device()
         self.n_gpu_layers = n_gpu_layers
+        self.n_ctx = n_ctx
         self._user_specified_dtype = torch_dtype  # track explicit user choice
         self.torch_dtype = self._resolve_torch_dtype(torch_dtype)
         self.load_in_8bit = load_in_8bit
@@ -734,6 +736,13 @@ class HorusProvider(BaseLLMProvider):
         info = getattr(self, "_model_info", None) or self.HORUS_MODELS.get(self.model_id, {})
         return int(info.get("context_window", HORUS_CONTEXT_WINDOW))
 
+    def _runtime_context_window(self) -> int:
+        if self.n_ctx is not None:
+            return max(int(self.n_ctx), 1)
+        if self._is_gguf_model_id(self.model_id):
+            return min(self._context_window(), HORUS_CONTEXT_WINDOW)
+        return self._context_window()
+
     def load(self) -> "HorusProvider":
         self._configure_external_logging()
         if self.model is not None:
@@ -1161,7 +1170,7 @@ class HorusProvider(BaseLLMProvider):
 
         llama_kwargs: Dict[str, Any] = {
             "model_path": model_path,
-            "n_ctx": self._context_window(),
+            "n_ctx": self._runtime_context_window(),
             "verbose": False,
             "n_gpu_layers": self._resolve_n_gpu_layers(),
         }
@@ -1991,6 +2000,7 @@ class HorusProvider(BaseLLMProvider):
             "flash_attention": self.use_flash_attention,
             "attn_implementation": self.attn_implementation,
             "context_window": self._context_window(),
+            "runtime_context_window": self._runtime_context_window(),
             "chat_template": UNIFIED_CHAT_TEMPLATE_NAME,
             "tts_enabled": self.enable_tts,
             "tts_voice_id": self.tts_voice_id,