docs: update for v1.0.0

docs/docs/instructions/cli.md

@@ -542,12 +542,18 @@ Example:
 
 Use the FLM benchmarking tool to measure a model's performance across different context lengths.
 
-Each benchmark tests context lengths from `1k` to `32k`, running `8` iterations at each length.
+Each benchmark tests context lengths from `1k` to `32k`, running `2` iterations at each length.
 
 ```shell
 flm bench llama3.2:1b
 ```
 
+Change the iteration times by `bench-iteratons`:
+
+```shell
+flm bench llama3.2:1b --bench-iteratons 4
+```
+
 FLM prints the results in your terminal and also saves them as a CSV file in the current folder for later reference.
 
 ```text

docs/docs/models/SmolVLA.md

@@ -0,0 +1,25 @@
+---
+layout: docs
+title: SmolVLA
+nav_order: 12
+parent: Models
+---
+
+## 🧩 Model Card: [SmolVLA](https://huggingface.co/lerobot/smolvla_base)
+
+- **Type:** Vision-Language-Action
+- **Think:** No
+- **Tool Calling Support:** No  
+- **Base Model:** [lerobot/smolvla_base](https://huggingface.co/lerobot/smolvla_base)
+- **Quantization:** bf16
+- **Max Context Length:** 48 tokens  
+- **Max Camera Input:** 3 images
+
+📝 **Note:**
+
+- SmolVLA is a robotics policy model that maps camera images and language instructions directly to robot actions — it does not run in FLM's standard CLI or Server chat modes.
+- For detailed usage instructions, please refer to:
+    - HuggingFace: [FastFlowLM/smolvla](https://huggingface.co/FastFlowLM/smolvla)
+    - ModelScope: [amd/smolvla](https://modelscope.cn/models/amd/smolvla)
+
+---

docs/docs/models/index.md

@@ -26,4 +26,5 @@ has_children: true
 - [Phi](phi/)
 - [Nanbeige](nanbeige/)
 - [Whisper](whisper/)
-- [Embedding Gemma](embeddinggemma/)
+- [Embedding Gemma](embeddinggemma/)
+- [SmolVLA](smolvla/)