change public func decode's behavior

Sources/LLM/LLM.swift

@@ -77,7 +77,7 @@ open class LLM: ObservableObject {
         let processorCount = UInt32(ProcessInfo().processorCount)
         self.maxTokenCount = Int(min(maxTokenCount, llama_n_ctx_train(model)))
         params.seed = seed
-        params.n_ctx = UInt32(maxTokenCount) + (maxTokenCount % 2 == 1 ? 1 : 2)
+        params.n_ctx = UInt32(self.maxTokenCount)
         params.n_batch = params.n_ctx
         params.n_threads = processorCount
         params.n_threads_batch = processorCount
@@ -359,10 +359,14 @@ open class LLM: ObservableObject {
     }
 
     private var multibyteCharacter: [CUnsignedChar] = []
-    public func decode(_ token: Token) -> String {
+    private func decode(_ token: Token) -> String {
         return model.decode(token, with: &multibyteCharacter)
     }
 
+    public func decode(_ tokens: [Token]) -> String {
+        return tokens.map({model.decodeOnly($0)}).joined()
+    }
+
     @inlinable
     public func encode(_ text: borrowing String) -> [Token] {
         model.encode(text)

Tests/LLMTests/LLMTests.swift

@@ -219,4 +219,13 @@ final class LLMTests: XCTestCase {
         await bot.respond(to: input)
         #assert(bot.output == "tl;dr")
     }
+
+    func testEncodingDecodingFromHuggingFaceModel() async throws {
+        let bot = try await LLM(from: model)
+        let input = "have you heard of this so-called LLM.swift library?"
+        let tokens = bot.encode(input)
+        let decoded = bot.decode(tokens).trimmingCharacters(in: .whitespacesAndNewlines)
+        #assert(!tokens.isEmpty)
+        #assert(decoded == input)
+    }
 }