refactor(llms): Make all LLM options fields nullable and add copyWith (…

…#284)

analysis_options.yaml

@@ -30,7 +30,7 @@ linter:
     - avoid_null_checks_in_equality_operators
     - avoid_positional_boolean_parameters
     - avoid_print
-    - avoid_redundant_argument_values
+    # - avoid_redundant_argument_values # Sometimes is useful to be explicit
     - avoid_relative_lib_imports
     - avoid_renaming_method_parameters
     - avoid_return_types_on_setters

docs/expression_language/cookbook/tools.md

@@ -4,10 +4,7 @@ Tools are also runnables, and can therefore be used within a chain:
 
 ```dart
 final openaiApiKey = Platform.environment['OPENAI_API_KEY'];
-final model = ChatOpenAI(
-  apiKey: openaiApiKey,
-  defaultOptions: const ChatOpenAIOptions(temperature: 0),
-);
+final model = ChatOpenAI(apiKey: openaiApiKey);
 const stringOutputParser = StringOutputParser();
 
 final promptTemplate = ChatPromptTemplate.fromTemplate('''
@@ -29,6 +26,7 @@ final chain = Runnable.getMapFromInput() |
 final res = await chain.invoke(
   'If I had 3 apples and you had 5 apples but we ate 3. '
   'If we cut the remaining apples in half, how many pieces would we have?',
+  options: const ChatOpenAIOptions(temperature: 0),
 );
 print(res);
 // 10.0

examples/browser_summarizer/pubspec.lock

@@ -465,10 +465,10 @@ packages:
     dependency: transitive
     description:
       name: uuid
-      sha256: "22c94e5ad1e75f9934b766b53c742572ee2677c56bc871d850a57dad0f82127f"
+      sha256: "8c951c9cb6504b2aa6b3666e6de504032d9baec24bf4cbabd3eea9edd73d4d77"
       url: "https://pub.dev"
     source: hosted
-    version: "4.2.2"
+    version: "4.3.2"
   vector_math:
     dependency: transitive
     description:

examples/docs_examples/bin/expression_language/cookbook/tools.dart

@@ -10,12 +10,7 @@ void main(final List<String> arguments) async {
 
 Future<void> _calculator() async {
   final openaiApiKey = Platform.environment['OPENAI_API_KEY'];
-  final model = ChatOpenAI(
-    apiKey: openaiApiKey,
-    defaultOptions: const ChatOpenAIOptions(
-      temperature: 0,
-    ),
-  );
+  final model = ChatOpenAI(apiKey: openaiApiKey);
   const stringOutputParser = StringOutputParser();
 
   final promptTemplate = ChatPromptTemplate.fromTemplate('''
@@ -37,6 +32,7 @@ MATH EXPRESSION:''');
   final res = await chain.invoke(
     'If I had 3 apples and you had 5 apples but we ate 3. '
     'If we cut the remaining apples in half, how many pieces would we have?',
+    options: const ChatOpenAIOptions(temperature: 0),
   );
   print(res);
   // 10.0

examples/docs_examples/pubspec.lock

@@ -370,10 +370,10 @@ packages:
     dependency: transitive
     description:
       name: uuid
-      sha256: bb55f38968b9427ce5dcdb8aaaa41049282195e0cfa4cf48593572fa3d1f36bc
+      sha256: "8c951c9cb6504b2aa6b3666e6de504032d9baec24bf4cbabd3eea9edd73d4d77"
       url: "https://pub.dev"
     source: hosted
-    version: "4.3.1"
+    version: "4.3.2"
   vector_math:
     dependency: transitive
     description:

examples/hello_world_backend/pubspec.lock

@@ -306,10 +306,10 @@ packages:
     dependency: transitive
     description:
       name: uuid
-      sha256: bb55f38968b9427ce5dcdb8aaaa41049282195e0cfa4cf48593572fa3d1f36bc
+      sha256: "8c951c9cb6504b2aa6b3666e6de504032d9baec24bf4cbabd3eea9edd73d4d77"
       url: "https://pub.dev"
     source: hosted
-    version: "4.3.1"
+    version: "4.3.2"
   vector_math:
     dependency: transitive
     description:

examples/hello_world_cli/pubspec.lock

@@ -266,10 +266,10 @@ packages:
     dependency: transitive
     description:
       name: uuid
-      sha256: bb55f38968b9427ce5dcdb8aaaa41049282195e0cfa4cf48593572fa3d1f36bc
+      sha256: "8c951c9cb6504b2aa6b3666e6de504032d9baec24bf4cbabd3eea9edd73d4d77"
       url: "https://pub.dev"
     source: hosted
-    version: "4.3.1"
+    version: "4.3.2"
   vector_math:
     dependency: transitive
     description:

examples/hello_world_flutter/pubspec.lock

@@ -324,10 +324,10 @@ packages:
     dependency: transitive
     description:
       name: uuid
-      sha256: "22c94e5ad1e75f9934b766b53c742572ee2677c56bc871d850a57dad0f82127f"
+      sha256: "8c951c9cb6504b2aa6b3666e6de504032d9baec24bf4cbabd3eea9edd73d4d77"
       url: "https://pub.dev"
     source: hosted
-    version: "4.2.2"
+    version: "4.3.2"
   vector_math:
     dependency: transitive
     description:

packages/googleai_dart/pubspec.yaml

@@ -27,7 +27,7 @@ dev_dependencies:
   build_runner: ^2.4.6
   freezed: ^2.4.5
   json_serializable: ^6.7.1
-  #  openapi_spec: ^0.7.8
+#  openapi_spec: ^0.7.8
   openapi_spec:
     git:
       url: https://github.com/davidmigloz/openapi_spec.git

packages/langchain/lib/src/model_io/language_models/base.dart

@@ -1,3 +1,5 @@
+import 'package:meta/meta.dart';
+
 import '../../core/core.dart';
 import '../chat_models/models/models.dart';
 import '../prompts/models/models.dart';
@@ -83,4 +85,33 @@ abstract class BaseLanguageModel<Input extends Object,
 
   @override
   String toString() => modelType;
+
+  /// Throws an error if the model id is not specified.
+  @protected
+  Never throwNullModelError() {
+    throw ArgumentError('''
+Null model in $runtimeType.
+        
+You need to specify the id of model to use either in `$runtimeType.defaultOptions` 
+or in the options passed when invoking the model.
+
+Example:
+```
+// In defaultOptions
+final model = $runtimeType(
+  defaultOptions: ${runtimeType}Options(
+    model: 'model-id',
+  ),
+);
+
+// Or when invoking the model
+final res = await model.invoke(
+  prompt,
+  options: ${runtimeType}Options(
+    model: 'model-id',
+  ),
+);
+```
+''');
+  }
 }

packages/langchain_google/lib/src/chat_models/google_ai/chat_google_generative_ai.dart

@@ -146,7 +146,9 @@ class ChatGoogleGenerativeAI
     final Map<String, String>? headers,
     final Map<String, dynamic>? queryParams,
     final http.Client? client,
-    this.defaultOptions = const ChatGoogleGenerativeAIOptions(),
+    this.defaultOptions = const ChatGoogleGenerativeAIOptions(
+      model: 'gemini-pro',
+    ),
   }) : _client = GoogleAIClient(
           apiKey: apiKey,
           baseUrl: baseUrl,
@@ -173,7 +175,8 @@ class ChatGoogleGenerativeAI
     final ChatGoogleGenerativeAIOptions? options,
   }) async {
     final id = _uuid.v4();
-    final model = options?.model ?? defaultOptions.model;
+    final model =
+        options?.model ?? defaultOptions.model ?? throwNullModelError();
     final completion = await _client.generateContent(
       modelId: model,
       request: _generateCompletionRequest(messages, options: options),
@@ -237,7 +240,7 @@ class ChatGoogleGenerativeAI
     final ChatGoogleGenerativeAIOptions? options,
   }) async {
     final tokens = await _client.countTokens(
-      modelId: options?.model ?? defaultOptions.model,
+      modelId: options?.model ?? defaultOptions.model ?? throwNullModelError(),
       request: CountTokensRequest(
         contents: promptValue.toChatMessages().toContentList(),
       ),

packages/langchain_google/lib/src/chat_models/google_ai/models/models.dart

@@ -19,7 +19,7 @@ class ChatGoogleGenerativeAIOptions extends ChatModelOptions {
   /// The LLM to use.
   ///
   /// You can find a list of available models here: https://ai.google.dev/models
-  final String model;
+  final String? model;
 
   /// The maximum cumulative probability of tokens to consider when sampling.
   /// The model uses combined Top-k and nucleus sampling. Tokens are sorted
@@ -76,6 +76,30 @@ class ChatGoogleGenerativeAIOptions extends ChatModelOptions {
   /// is no safety setting for a given category provided in the list, the API will use
   /// the default safety setting for that category.
   final List<ChatGoogleGenerativeAISafetySetting>? safetySettings;
+
+  /// Creates a copy of this [ChatGoogleGenerativeAIOptions] object with the given fields
+  /// replaced with the new values.
+  ChatGoogleGenerativeAIOptions copyWith({
+    final String? model,
+    final double? topP,
+    final int? topK,
+    final int? candidateCount,
+    final int? maxOutputTokens,
+    final double? temperature,
+    final List<String>? stopSequences,
+    final List<ChatGoogleGenerativeAISafetySetting>? safetySettings,
+  }) {
+    return ChatGoogleGenerativeAIOptions(
+      model: model ?? this.model,
+      topP: topP ?? this.topP,
+      topK: topK ?? this.topK,
+      candidateCount: candidateCount ?? this.candidateCount,
+      maxOutputTokens: maxOutputTokens ?? this.maxOutputTokens,
+      temperature: temperature ?? this.temperature,
+      stopSequences: stopSequences ?? this.stopSequences,
+      safetySettings: safetySettings ?? this.safetySettings,
+    );
+  }
 }
 
 /// {@template chat_google_generative_ai_safety_setting}

packages/langchain_google/lib/src/chat_models/vertex_ai/chat_vertex_ai.dart

@@ -116,7 +116,10 @@ class ChatVertexAI extends BaseChatModel<ChatVertexAIOptions> {
     required final String project,
     final String location = 'us-central1',
     final String? rootUrl,
-    this.defaultOptions = const ChatVertexAIOptions(),
+    this.defaultOptions = const ChatVertexAIOptions(
+      publisher: 'google',
+      model: 'chat-bison',
+    ),
   }) : client = VertexAIGenAIClient(
           httpClient: httpClient,
           project: project,
@@ -158,25 +161,33 @@ class ChatVertexAI extends BaseChatModel<ChatVertexAIOptions> {
     final examples = (options?.examples ?? defaultOptions.examples)
         ?.map((final e) => e.toVertexAIChatExample())
         .toList(growable: false);
+    final model =
+        options?.model ?? defaultOptions.model ?? throwNullModelError();
 
     final result = await client.chat.predict(
       context: context,
       examples: examples,
       messages: vertexMessages,
-      publisher: options?.publisher ?? defaultOptions.publisher,
-      model: options?.model ?? defaultOptions.model,
+      publisher: options?.publisher ??
+          defaultOptions.publisher ??
+          ArgumentError.checkNotNull(
+            defaultOptions.publisher,
+            'VertexAIOptions.publisher',
+          ),
+      model: model,
       parameters: VertexAITextChatModelRequestParams(
         maxOutputTokens:
-            options?.maxOutputTokens ?? defaultOptions.maxOutputTokens,
-        temperature: options?.temperature ?? defaultOptions.temperature,
-        topP: options?.topP ?? defaultOptions.topP,
-        topK: options?.topK ?? defaultOptions.topK,
-        stopSequences: options?.stopSequences ?? defaultOptions.stopSequences,
+            options?.maxOutputTokens ?? defaultOptions.maxOutputTokens ?? 1024,
+        temperature: options?.temperature ?? defaultOptions.temperature ?? 0.2,
+        topP: options?.topP ?? defaultOptions.topP ?? 0.95,
+        topK: options?.topK ?? defaultOptions.topK ?? 40,
+        stopSequences:
+            options?.stopSequences ?? defaultOptions.stopSequences ?? const [],
         candidateCount:
-            options?.candidateCount ?? defaultOptions.candidateCount,
+            options?.candidateCount ?? defaultOptions.candidateCount ?? 1,
       ),
     );
-    return result.toChatResult(id, options?.model ?? defaultOptions.model);
+    return result.toChatResult(id, model);
   }
 
   /// Tokenizes the given prompt using tiktoken.