diff --git a/docs/references/sampling_params.md b/docs/references/sampling_params.md
index 77d7c9f82e7..013dfc40963 100644
--- a/docs/references/sampling_params.md
+++ b/docs/references/sampling_params.md
@@ -3,52 +3,63 @@ This doc describes the sampling parameters of the SGLang Runtime.
 It is the low-level endpoint of the runtime.
 If you want a high-level endpoint that can automatically handle chat templates, consider using the [OpenAI Compatible API](../backend/openai_api_completions.ipynb).
 
-The `/generate` endpoint accepts the following arguments in the JSON format.
+The `/generate` endpoint accepts the following arguments in JSON format. These parameters allow customization of input data, output control, and configuration options.
 
 ```python
 @dataclass
 class GenerateReqInput:
-    # The input prompt. It can be a single prompt or a batch of prompts.
+    """Defines the input parameters for the /generate endpoint in SGLang."""
+    # The input prompt, can be a single string or a batch of strings.
     text: Optional[Union[List[str], str]] = None
-    # The token ids for text; one can specify either text or input_ids
+    # The token IDs for the input prompt; can specify either `text` or `input_ids`.
     input_ids: Optional[Union[List[List[int]], List[int]]] = None
-    # The embeddings for input_ids; one can specify either text or input_ids or input_embeds.
+    # The embeddings for the input IDs; can specify either `text`, `input_ids`, or `input_embeds`.
     input_embeds: Optional[Union[List[List[List[float]]], List[List[float]]]] = None
-    # The image input. It can be a file name, a url, or base64 encoded string.
-    # See also python/sglang/srt/utils.py:load_image.
+    # Image input, can be a file name, a URL, or a base64-encoded string.
+    # Refer to python/sglang/srt/utils.py:load_image for details.
     image_data: Optional[Union[List[str], str]] = None
-    # The sampling_params. See descriptions below.
+
+    # Sampling-related parameters
+    # Parameters that define the sampling behavior, described in detail below.
     sampling_params: Optional[Union[List[Dict], Dict]] = None
-    # The request id.
+
+    ## Metadata
+    # Request ID to track the request (can be a single ID or a batch of IDs).
     rid: Optional[Union[List[str], str]] = None
-    # Whether to return logprobs.
+    # Whether to return log probabilities for generated tokens.
     return_logprob: Optional[Union[List[bool], bool]] = None
-    # If return logprobs, the start location in the prompt for returning logprobs.
-    # By default, this value is "-1", which means it will only return logprobs for output tokens.
+    # The starting position in the prompt for returning log probabilities.
+    # Defaults to "-1", which means it only returns log probabilities for output tokens.
     logprob_start_len: Optional[Union[List[int], int]] = None
-    # If return logprobs, the number of top logprobs to return at each position.
+    # The number of top log probabilities to return for each token position.
     top_logprobs_num: Optional[Union[List[int], int]] = None
-    # Whether to detokenize tokens in text in the returned logprobs.
+    # Whether to include detokenized text in the returned log probabilities.
     return_text_in_logprobs: bool = False
-    # Whether to stream output.
+
+    ## Output streaming and logging
+    # Whether to stream the output in real-time.
     stream: bool = False
-    # Whether to log metrics for this request (e.g. health_generate calls do not log metrics)
+    # Whether to log metrics for this request (e.g., for health checks, metrics logging is disabled).
     log_metrics: bool = True
 
-    # The modalities of the image data [image, multi-images, video]
+    ## Multi-modal and session-related parameters
+    # Defines the modalities for image input (e.g., `image`, `multi-images`, or `video`).
     modalities: Optional[List[str]] = None
-    # LoRA related
+    # LoRA-related parameters for low-rank adaptation models.
     lora_path: Optional[Union[List[Optional[str]], Optional[str]]] = None
 
-    # Session info for continual prompting
+    ## Session and custom logit processor
+    # Session parameters for continual prompting across multiple requests.
     session_params: Optional[Union[List[Dict], Dict]] = None
-    # Custom logit processor for advanced sampling control. Must be a serialized instance
-    # of `CustomLogitProcessor` in python/sglang/srt/sampling/custom_logit_processor.py
-    # Use the processor's `to_str()` method to generate the serialized string.
+    # Custom logit processor for advanced sampling control.
+    # Must be a serialized instance of `CustomLogitProcessor` from
+    # python/sglang/srt/sampling/custom_logit_processor.py.
+    # Use the processor's `to_str()` method for serialization.
     custom_logit_processor: Optional[Union[List[Optional[str]], str]] = None
+
 ```
 
-The `sampling_params` follows this format
+The `sampling_params` configure the generation process, including sampling strategies and output constraints. Below is a detailed breakdown of each parameter.
 
 ```python
 # The maximum number of output tokens