chat method

@override Future<String> chat(List<Map<String, String>> messages, { int maxTokens = 256, })

Generates a response with conversation history context.

messages represents the sequence of conversational exchanges.

Implementation

@override
Future<String> chat(List<Map<String, String>> messages, {int maxTokens = 256}) async {
  if (!isReady) {
    throw Exception('LocalLlamaProvider: Local LLM is not loaded. Call initialize() first.');
  }

  try {
    // Reconstruct prompt in standard ChatML/Llama chat template format
    final formattedPrompt = _formatChatTemplate(messages);

    return await generate(formattedPrompt, maxTokens: maxTokens);
  } catch (e) {
    SintSentinel.logger.e('LocalLlamaProvider: Chat inference failed: $e');
    rethrow;
  }
}