Set pad_token_id in ModelConfig as Option

It can be provided / set to `null` e.g.,
https://huggingface.co/jinaai/jina-embeddings-v5-text-small-retrieval/blob/main/config.json#L51

Author: alvarobartt

router/src/lib.rs

@@ -108,6 +108,7 @@ pub async fn run(
     // Load config
     let config_path = model_root.join("config.json");
     let config = fs::read_to_string(config_path).context("`config.json` not found")?;
+    tracing::info!("config contains {:?}", config);
     let config: ModelConfig =
         serde_json::from_str(&config).context("Failed to parse `config.json`")?;
 
@@ -181,7 +182,7 @@ pub async fn run(
         || &config.model_type == "camembert"
         || &config.model_type == "roberta"
     {
-        config.pad_token_id + 1
+        config.pad_token_id.unwrap_or(0) + 1
     } else {
         0
     };
@@ -459,7 +460,7 @@ pub struct ModelConfig {
     #[serde(alias = "n_positions")]
     pub max_position_embeddings: usize,
     #[serde(default)]
-    pub pad_token_id: usize,
+    pub pad_token_id: Option<usize>,
     pub id2label: Option<HashMap<String, String>>,
     pub label2id: Option<HashMap<String, usize>>,
     pub auto_map: Option<HashMap<String, String>>,