docs: Add initial fern docs (#676)

Author: debermudez

.cursor/skills/docs-to-fern/SKILL_md

@@ -1,7 +1,8 @@
-<!--
+---
 # SPDX-FileCopyrightText: Copyright (c) 2025-2026 NVIDIA CORPORATION & AFFILIATES. All rights reserved.
 # SPDX-License-Identifier: Apache-2.0
--->
+sidebar-title: Prefix Synthesis API Reference
+---
 # Prefix Synthesis API Reference
 
 Complete API documentation for the prefix synthesis module.
@@ -556,4 +557,4 @@ except ValueError as e:
 ## See Also
 
 - [Prefix Synthesis Tutorial](../tutorials/prefix-synthesis.md)
-- [Trace Replay](../benchmark_modes/trace_replay.md)
+- [Trace Replay](../benchmark-modes/trace-replay.md)

docs/api/synthesis.md

@@ -1,7 +1,8 @@
-<!--
-SPDX-FileCopyrightText: Copyright (c) 2025-2026 NVIDIA CORPORATION & AFFILIATES. All rights reserved.
-SPDX-License-Identifier: Apache-2.0
--->
+---
+# SPDX-FileCopyrightText: Copyright (c) 2025-2026 NVIDIA CORPORATION & AFFILIATES. All rights reserved.
+# SPDX-License-Identifier: Apache-2.0
+sidebar-title: Architecture of AIPerf
+---
 
 # Architecture of AIPerf
 

docs/architecture.md

@@ -1,7 +1,8 @@
-<!--
-SPDX-FileCopyrightText: Copyright (c) 2024-2025 NVIDIA CORPORATION & AFFILIATES. All rights reserved.
-SPDX-License-Identifier: Apache-2.0
--->
+---
+# SPDX-FileCopyrightText: Copyright (c) 2024-2025 NVIDIA CORPORATION & AFFILIATES. All rights reserved.
+# SPDX-License-Identifier: Apache-2.0
+sidebar-title: Benchmark Datasets
+---
 
 This document describes datasets that AIPerf can use to generate stimulus. Additional support is under development, so check back often.
 
@@ -39,12 +40,12 @@ This document describes datasets that AIPerf can use to generate stimulus. Addit
     <tr>
     <td><strong>Mooncake</strong></td>
     <td style="text-align: center;">✅</td>
-    <td>Mooncake trace file <a href="benchmark_modes/trace_replay.md"><code>--input-file your_trace_file.jsonl --custom-dataset-type mooncake_trace</code></a></td>
+    <td>Mooncake trace file <a href="benchmark-modes/trace-replay.md"><code>--input-file your_trace_file.jsonl --custom-dataset-type mooncake_trace</code></a></td>
     </tr>
     <tr>
       <td><strong>ShareGPT</strong></td>
       <td style="text-align: center;">✅</td>
-      <td>Conversations from <a href="https://huggingface.co/datasets/anon8231489123/ShareGPT_Vicuna_unfiltered/resolve/main/ShareGPT_V3_unfiltered_cleaned_split.json"><code>--public-dataset sharegpt</code></td>
+      <td>Conversations from <a href="https://huggingface.co/datasets/anon8231489123/ShareGPT_Vicuna_unfiltered/resolve/main/ShareGPT_V3_unfiltered_cleaned_split.json"><code>--public-dataset sharegpt</code></a></td>
     </tr>
   </tbody>
 </table>

docs/benchmark_datasets.md docs/benchmark-datasets.mddocs/benchmark_datasets.md renamed to docs/benchmark-datasets.md

@@ -1,7 +1,8 @@
-<!--
+---
 # SPDX-FileCopyrightText: Copyright (c) 2025-2026 NVIDIA CORPORATION & AFFILIATES. All rights reserved.
 # SPDX-License-Identifier: Apache-2.0
--->
+sidebar-title: Load Generator Options Reference
+---
 # Load Generator Options Reference
 
 This guide provides a comprehensive reference for all load generator CLI options in AIPerf, including a compatibility matrix showing which options work together.

…enchmark_modes/timing-modes-reference.md …enchmark-modes/timing-modes-reference.mddocs/benchmark_modes/timing-modes-reference.md renamed to docs/benchmark-modes/timing-modes-reference.md docs/benchmark_modes/timing-modes-reference.md renamed to docs/benchmark-modes/timing-modes-reference.md

@@ -1,7 +1,8 @@
-<!--
-SPDX-FileCopyrightText: Copyright (c) 2025-2026 NVIDIA CORPORATION & AFFILIATES. All rights reserved.
-SPDX-License-Identifier: Apache-2.0
--->
+---
+# SPDX-FileCopyrightText: Copyright (c) 2025-2026 NVIDIA CORPORATION & AFFILIATES. All rights reserved.
+# SPDX-License-Identifier: Apache-2.0
+sidebar-title: Trace Replay with Mooncake Traces
+---
 
 # Trace Replay with Mooncake Traces
 
@@ -61,18 +62,18 @@ Example entry:
 
 Create a trace file with timing information:
 
-<!-- aiperf-run-vllm-default-openai-endpoint-server -->
+{/* aiperf-run-vllm-default-openai-endpoint-server */}
 ```bash
 cat > custom_trace.jsonl << 'EOF'
 {"timestamp": 0, "input_length": 1200, "output_length": 52, "hash_ids": [0, 1, 2]}
 {"timestamp": 105, "input_length": 1800, "output_length": 26, "hash_ids": [0, 3, 4, 5]}
 {"timestamp": 274, "input_length": 1300, "output_length": 52, "hash_ids": [1, 4, 6]}
 EOF
 ```
-<!-- /aiperf-run-vllm-default-openai-endpoint-server -->
+{/* /aiperf-run-vllm-default-openai-endpoint-server */}
 Run AIPerf with the trace file:
 
-<!-- aiperf-run-vllm-default-openai-endpoint-server -->
+{/* aiperf-run-vllm-default-openai-endpoint-server */}
 ```bash
 aiperf profile \
     --model Qwen/Qwen3-0.6B \
@@ -83,7 +84,7 @@ aiperf profile \
     --custom-dataset-type mooncake_trace \
     --fixed-schedule
 ```
-<!-- /aiperf-run-vllm-default-openai-endpoint-server -->
+{/* /aiperf-run-vllm-default-openai-endpoint-server */}
 
 The `--fixed-schedule` flag tells AIPerf to send requests at the exact timestamps specified in the trace. This reproduces the original timing pattern.
 
@@ -114,7 +115,7 @@ The `tools` field is only valid when `messages` is provided. It is injected dire
 
 For real-world benchmarking, use the FAST25 production trace data from the Mooncake research paper:
 
-<!-- aiperf-run-vllm-default-openai-endpoint-server -->
+{/* aiperf-run-vllm-default-openai-endpoint-server */}
 ```bash
 # Download the Mooncake trace data
 curl -Lo mooncake_trace.jsonl https://raw.githubusercontent.com/kvcache-ai/Mooncake/refs/heads/main/FAST25-release/arxiv-trace/mooncake_trace.jsonl
@@ -132,4 +133,4 @@ aiperf profile \
     --custom-dataset-type mooncake_trace \
     --fixed-schedule
 ```
-<!-- /aiperf-run-vllm-default-openai-endpoint-server -->
+{/* /aiperf-run-vllm-default-openai-endpoint-server */}