@umituz/react-native-ai-fal-provider 2.0.14 → 2.0.16

package/src/domain/constants/models/text-to-image.README.md

@@ -1,237 +0,0 @@
-# Text-to-Image Generation
-
-Generates high-quality images from text-based prompts.
-
-**Location:** `src/domain/constants/models/text-to-image.ts`
-
-## Overview
-
-Text-to-image generation creates visual content from textual descriptions using AI models. This module provides model configurations for various image generation capabilities with different quality levels and cost structures.
-
-## Purpose
-
-Provides image generation models by:
-- Supporting multiple quality tiers (fast, quality, professional)
-- Offering flexible image sizes and aspect ratios
-- Enabling batch image generation
-- Providing safety checking
-- Supporting configurable inference steps
-
-## Import
-
-```typescript
-import {
-  falProvider,
-  useFalGeneration
-} from '@umituz/react-native-ai-fal-provider';
-```
-
-## Available Models
-
-### Flux Schnell
-- **Model ID:** `fal-ai/flux/schnell`
-- **Cost:** 1 credit (Free), 0.5 credit (Premium)
-- **Description:** Fast and efficient image generation (Default)
-- **Use Cases:** Quick iterations, drafts, preview generations
-
-### Flux Dev
-- **Model ID:** `fal-ai/flux/dev`
-- **Cost:** 2 credits (Free), 1 credit (Premium)
-- **Description:** High-quality image generation
-- **Use Cases:** Production images, enhanced quality
-
-### Flux Pro
-- **Model ID:** `fal-ai/flux-pro`
-- **Cost:** 3 credits (Free), 1.5 credits (Premium)
-- **Description:** Professional-level image generation
-- **Use Cases:** Professional work, highest quality requirements
-
-## Parameters
-
-### Common Parameters
-
-| Parameter | Type | Description | Default |
-|-----------|------|-------------|---------|
-| `prompt` | `string` | Description of image to generate | - |
-| `image_size` | `string` | Output image dimensions | `square_hd` |
-| `num_inference_steps` | `number` | Number of inference steps | 4 |
-| `num_images` | `number` | Number of images to generate | 1 |
-| `enable_safety_checker` | `boolean` | Enable safety checking | true |
-
-### Image Sizes
-
-- `square_hd`: 1024x1024
-- `square`: 512x512
-- `portrait_4_3`: 832x1104
-- `portrait_16_9`: 832x1216
-- `landscape_4_3`: 1104x832
-- `landscape_16_9`: 1216x832
-
-## Usage Guidelines
-
-### For Image Generation
-
-**Generation Pattern:**
-1. Select model based on quality needs
-2. Construct descriptive prompt
-3. Set appropriate image size
-4. Configure inference steps (quality vs speed)
-5. Handle generation result
-
-**Best Practices:**
-- Use descriptive, detailed prompts
-- Match image size to intended use
-- Start with Schnell for quick iterations
-- Upgrade to Dev/Pro for final output
-- Enable safety checker for production
-
-### For Model Selection
-
-**Selection Pattern:**
-1. Evaluate quality requirements
-2. Consider cost constraints
-3. Balance speed vs quality
-4. Use appropriate model for task
-5. Scale up as needed
-
-**Model Choice Guidelines:**
-- Schnell: Testing, drafts, previews
-- Dev: Standard production use
-- Pro: Professional, highest quality
-
-## Best Practices
-
-### 1. Write Effective Prompts
-
-Create detailed descriptions:
-- Be specific about subjects and style
-- Include lighting and atmosphere details
-- Specify composition and framing
-- Mention mood and emotion
-- Use natural language descriptions
-
-### 2. Choose Right Model
-
-Select appropriate quality tier:
-- Use Schnell for rapid iteration
-- Upgrade to Dev for better quality
-- Use Pro for professional work
-- Consider cost vs quality trade-off
-- Test with lower tier first
-
-### 3. Optimize Parameters
-
-Balance quality and speed:
-- Lower steps for faster generation
-- Higher steps for better quality
-- Adjust image size for use case
-- Generate multiple images when needed
-- Enable safety checking
-
-### 4. Handle Results
-
-Process generated images:
-- Validate image quality
-- Check safety filter results
-- Handle generation failures
-- Display images appropriately
-- Store results efficiently
-
-### 5. Manage Costs
-
-Control generation costs:
-- Start with lower tier models
-- Limit batch generation
-- Use appropriate image sizes
-- Cache successful generations
-- Monitor credit usage
-
-## For AI Agents
-
-### When Using Text-to-Image Models
-
-**DO:**
-- Select appropriate model for task
-- Write detailed descriptive prompts
-- Set correct image sizes
-- Enable safety checking
-- Handle generation errors
-- Monitor credit usage
-- Follow parameter constraints
-
-**DON'T:**
-- Use highest tier for testing
-- Write vague or short prompts
-- Ignore safety checking
-- Forget to handle failures
-- Generate unnecessary batches
-- Exceed credit limits
-- Use wrong aspect ratios
-
-### When Writing Prompts
-
-**DO:**
-- Be specific and detailed
-- Include style information
-- Describe lighting and mood
-- Specify composition
-- Use natural language
-- Include important details
-- Reference art styles if needed
-
-**DON'T:**
-- Use vague descriptions
-- Forget key details
-- Ignore style guidance
-- Skip composition details
-- Use overly complex language
-- Include contradictory elements
-- Exceed reasonable length
-
-### When Selecting Models
-
-**DO:**
-- Start with Schnell for testing
-- Upgrade to Dev for production
-- Use Pro for professional work
-- Consider cost implications
-- Balance quality vs speed
-- Test before final generation
-- Scale appropriately
-
-**DON'T:**
-- Always use highest tier
-- Ignore cost constraints
-- Skip testing phase
-- Use Pro for drafts
-- Forget quality requirements
-- Waste credits on iterations
-- Select model randomly
-
-## Implementation Notes
-
-**Location:** `src/domain/constants/models/text-to-image.ts`
-
-**Dependencies:**
-- FAL provider service
-- useFalGeneration hook
-- Image format utilities
-- Safety validators
-
-**Model Categories:**
-- Fast generation (Flux Schnell)
-- Quality generation (Flux Dev)
-- Professional generation (Flux Pro)
-
-**Import:**
-```typescript
-import {
-  falProvider,
-  useFalGeneration
-} from '@umituz/react-native-ai-fal-provider';
-```
-
-**Related:**
-- FAL provider: `src/infrastructure/services/fal-provider.ts`
-- Generation hook: `src/presentation/hooks/use-fal-generation.ts`
-- Default models: `src/domain/constants/default-models.constants.ts`

package/src/domain/constants/models/text-to-text.README.md

@@ -1,249 +0,0 @@
-# Text-to-Text Generation
-
-Generates text from text using LLM-based models.
-
-**Location:** `src/domain/constants/models/text-to-text.ts`
-
-## Overview
-
-Text-to-text generation uses Large Language Models (LLMs) to generate, transform, or analyze text content. This module provides model configurations for various text generation tasks with controllable parameters.
-
-## Purpose
-
-Provides text generation models by:
-- Supporting question answering
-- Enabling content generation
-- Facilitating text transformation
-- Providing summarization
-- Supporting translation tasks
-
-## Import
-
-```typescript
-import {
-  falProvider,
-  useFalGeneration
-} from '@umituz/react-native-ai-fal-provider';
-```
-
-## Available Models
-
-### Llama 3 8B Instruct
-- **Model ID:** `fal-ai/llama-3-8b-instruct`
-- **Cost:** 0.1 credits (Free), 0.05 credits (Premium)
-- **Description:** Fast and reliable text generation (Default)
-- **Use Cases:** QA, content generation, summarization, translation
-
-## Parameters
-
-### Basic Parameters
-
-**Configuration:**
-- `prompt`: Input prompt or question
-- `max_tokens`: Maximum tokens to generate (default: 512)
-- `temperature`: Creativity level 0.0 - 1.0 (default: 0.7)
-- `top_p`: Nucleus sampling 0.0 - 1.0 (default: 0.9)
-- `top_k`: Top-k sampling (default: 40)
-- `repetition_penalty`: Reduce repetitions 1.0 - 2.0
-
-### Parameter Descriptions
-
-| Parameter | Range | Description |
-|-----------|-------|-------------|
-| `prompt` | string | Input text or question |
-| `max_tokens` | 1 - 4096 | Maximum tokens to generate |
-| `temperature` | 0.0 - 1.0 | Low = focused, High = creative |
-| `top_p` | 0.0 - 1.0 | Nucleus sampling threshold |
-| `top_k` | 1 - 100 | Top-k sampling value |
-| `repetition_penalty` | 1.0 - 2.0 | Reduce repetitions |
-
-## Usage Guidelines
-
-### For Text Generation
-
-**Generation Pattern:**
-1. Construct clear prompt
-2. Set appropriate max_tokens
-3. Configure temperature for task
-4. Generate text
-5. Handle output appropriately
-
-**Best Practices:**
-- Write clear, specific prompts
-- Use system prompts for behavior
-- Adjust temperature for task type
-- Set appropriate max_tokens
-- Validate output quality
-
-### For Temperature Selection
-
-**Temperature Guidelines:**
-
-**Low (0.0 - 0.3):**
-- Factual responses
-- Code generation
-- Deterministic output
-- Technical content
-
-**Medium (0.4 - 0.7):**
-- Balanced output
-- General assistance
-- Standard responses
-- Most use cases
-
-**High (0.8 - 1.0):**
-- Creative writing
-- Diverse outputs
-- Brainstorming
-- Fiction generation
-
-## Best Practices
-
-### 1. Engineer Effective Prompts
-
-Create well-structured prompts:
-- Define role or context
-- Provide clear instructions
-- Specify output format
-- Include examples if needed
-- Set constraints explicitly
-
-**Prompt Structure:**
-```
-Role/Context
-Task/Instructions
-Input Content
-Output Format/Constraints
-```
-
-### 2. Optimize Token Usage
-
-Manage generation efficiently:
-- Set appropriate max_tokens for task
-- Use shorter prompts when possible
-- Limit output length for simple tasks
-- Monitor token consumption
-- Cache common responses
-
-### 3. Select Right Temperature
-
-Choose temperature based on task:
-- Factual QA: 0.2 - 0.3
-- Code generation: 0.2 - 0.4
-- Translation: 0.3 - 0.5
-- Summarization: 0.5 - 0.7
-- General assistance: 0.6 - 0.7
-- Creative writing: 0.8 - 1.0
-
-### 4. Use System Prompts
-
-Guide model behavior:
-- Define role explicitly
-- Set behavioral constraints
-- Specify response style
-- Include format requirements
-- Maintain consistency
-
-### 5. Handle Edge Cases
-
-Manage challenging scenarios:
-- Validate input prompts
-- Handle timeout errors
-- Check quota limits
-- Validate output quality
-- Implement fallback strategies
-
-## For AI Agents
-
-### When Using Text-to-Text Models
-
-**DO:**
-- Write clear specific prompts
-- Use system prompts for guidance
-- Set appropriate temperature
-- Configure max_tokens correctly
-- Validate output quality
-- Handle errors appropriately
-- Monitor token usage
-- Test with various prompts
-
-**DON'T:**
-- Write vague prompts
-- Ignore temperature settings
-- Set excessive max_tokens
-- Skip output validation
-- Forget error handling
-- Waste tokens on long outputs
-- Use wrong temperature
-- Assume perfect output
-
-### When Engineering Prompts
-
-**DO:**
-- Define role or context
-- Provide clear instructions
-- Specify output format
-- Include examples when helpful
-- Set explicit constraints
-- Test prompt variations
-- Iterate on prompt design
-
-**DON'T:**
-- Use vague instructions
-- Skip context setting
-- Forget format specification
-- Ignore constraint setting
-- Use overly complex prompts
-- Skip testing
-- Use first draft
-
-### When Configuring Parameters
-
-**DO:**
-- Match temperature to task
-- Set appropriate max_tokens
-- Use repetition penalty
-- Configure sampling parameters
-- Test different settings
-- Monitor output quality
-- Adjust based on results
-
-**DON'T:**
-- Use default temperature always
-- Set excessive max_tokens
-- Ignore sampling parameters
-- Skip parameter testing
-- Use arbitrary values
-- Forget output quality
-- Waste tokens
-
-## Implementation Notes
-
-**Location:** `src/domain/constants/models/text-to-text.ts`
-
-**Dependencies:**
-- FAL provider service
-- useFalGeneration hook
-- Text processing utilities
-- Prompt templates
-
-**Common Use Cases:**
-- Question answering
-- Content generation
-- Text summarization
-- Translation
-- Code generation
-- Creative writing
-
-**Import:**
-```typescript
-import {
-  falProvider,
-  useFalGeneration
-} from '@umituz/react-native-ai-fal-provider';
-```
-
-**Related:**
-- FAL provider: `src/infrastructure/services/fal-provider.ts`
-- Generation hook: `src/presentation/hooks/use-fal-generation.ts`
-- Default models: `src/domain/constants/default-models.constants.ts`