fix links

docs/docs/language/reference.md

@@ -34,7 +34,7 @@ LMQL's [current reference implementation](#reference-implementation) is written
 
 * The `lmql playground` offers an interactive interface for running, debugging and experimenting with LMQL programs. It is the recommended way to get started with LMQL. An online version of the playground is available at [lmql.ai/playground](https://lmql.ai/playground).
 
-* LMQL is available as a Python library, with the `lmql.query` function offering a simple way to use LMQL directly from within Python. For more information, please refer to the [Python API](#python-api) section.
+* LMQL is available as a Python library, with the `lmql.query` function offering a simple way to use LMQL directly from within Python. For more information, please refer to general [Documentation](overview.md).
 
 * The `lmql run` CLI tool can used to run LMQL programs directly from the command line.
 
@@ -44,9 +44,9 @@ For more information and example-based discussion on how to use LMQL, please ref
 
 The LMQL language comprises two syntax variants:
 
-* The modern, more minimalistic [standard syntax](#modern-syntax) that relies on a very small set of language constructs in an otherwise standard Python environment. This syntax is the main focus for the continued development of LMQL.
+* The modern, more minimalistic [standard syntax](#standard-syntax) that relies on a very small set of language constructs in an otherwise standard Python environment. This syntax is the main focus for the continued development of LMQL.
 
-* A *legacy* [standalone syntax](#legacy-syntax) that is more static in nature but relevant for standalone LMQL use-cases.
+* A *legacy* [standalone syntax](#standalone-syntax) that is more static in nature but relevant for standalone LMQL use-cases.
 
 Both syntax variants are compatible and can be used interchangeably. 
 

docs/docs/lib/generations.md

@@ -74,7 +74,7 @@ Generates a text completion based on a given prompt. Returns the full prompt + c
 
 **Return Value** The function returns a string or a list of strings, depending on the decoder in use (`decoder=argmax` yields a single sequence, `decoder="sample", n=2` yields two sequences, etc.).
 
-**Asynchronous** The function is asynchronous and should be used with [`asyncio`](https://docs.python.org/3/library/asyncio.html) and with `await`. When run in parallel, multiple generations will be batched and parallelized across multiple calls to the same model. For synchronous use, you can rely on [`LLM.generate_sync`](#llm-generate_sync), note however, that in this case, the function will block the current thread until generation is complete.
+**Asynchronous** The function is asynchronous and should be used with [`asyncio`](https://docs.python.org/3/library/asyncio.html) and with `await`. When run in parallel, multiple generations will be batched and parallelized across multiple calls to the same model. For synchronous use, you can rely on [`LLM.generate_sync`](#llm-generate-sync), note however, that in this case, the function will block the current thread until generation is complete.
 
 ### `LLM.generate_sync(...)`
 
@@ -134,7 +134,7 @@ async def lmql.generate(
 `lmql.generate` generates text completions based on a given prompt and behaves just like [`LLM.generate`](#llm-generate), 
 with the provided `model` instance or model name.
 
-If no `model` is provided, the default model is used. See [`lmql.set_default_model`](#lmql-set_default_model) for more information.
+If no `model` is provided, the default model is used. See [`lmql.set_default_model`](#lmql-set-default-model) for more information.
 
 ### `lmql.generate_sync(...)`
 
@@ -154,7 +154,7 @@ async def score(
 `lmql.score` scores different continuation `values` for a given `prompt` and behaves just like [`LLM.score`](#llm-score),
 with the provided `model` instance or model name.
 
-If no `model` is provided, the default model is used. See [`lmql.set_default_model`](#lmql-set_default_model) for more information.
+If no `model` is provided, the default model is used. See [`lmql.set_default_model`](#lmql-set-default-model) for more information.
 
 ### `lmql.score_sync(...)`
 

docs/docs/models/hf.md

@@ -7,7 +7,7 @@ LMQL relies on a two-process architecture: The inference process (long-running)
 
 This architecture is advantageous for locally-hosted models, as the model loading time can be quite long or the required GPU hardware might not even be available on the client machine. 
 
-This chapter first discusses how to use of the two-process inference API, and then presents an alternative on how to leverage [In-Process Model Loading](#in-process-model-loading), which avoids the need for a separate server process within the same architecture.
+This chapter first discusses how to use of the two-process inference API, and then presents an alternative on how to leverage [In-Process Model Loading](#in-process-models), which avoids the need for a separate server process within the same architecture.
 
 ![Inference Architecture](./inference.svg)