Merge branch 'main' into pre/beta

Author: VinciGit00

CHANGELOG.md

@@ -33,6 +33,8 @@
 
 * implement ScrapeGraph class for only web scraping automation ([612c644](https://github.com/ScrapeGraphAI/Scrapegraph-ai/commit/612c644623fa6f4fe77a64a5f1a6a4d6cd5f4254))
 * Implement SmartScraperMultiParseMergeFirstGraph class that scrapes a list of URLs and merge the content first and finally generates answers to a given prompt. ([3e3e1b2](https://github.com/ScrapeGraphAI/Scrapegraph-ai/commit/3e3e1b2f3ae8ed803d03b3b44b199e139baa68d4))
+=======
+## [1.26.7](https://github.com/ScrapeGraphAI/Scrapegraph-ai/compare/v1.26.6...v1.26.7) (2024-10-19)
 
 
 ### Bug Fixes
@@ -70,6 +72,8 @@
 * add conditional node structure to the smart_scraper_graph and implemented a structured way to check condition ([cacd9cd](https://github.com/ScrapeGraphAI/Scrapegraph-ai/commit/cacd9cde004dace1a7dcc27981245632a78b95f3))
 
 
+* removed tokenizer ([a184716](https://github.com/ScrapeGraphAI/Scrapegraph-ai/commit/a18471688f0b79f06fb7078b01b68eeddc88eae4))
+
 ## [1.26.6](https://github.com/ScrapeGraphAI/Scrapegraph-ai/compare/v1.26.5...v1.26.6) (2024-10-18)
 
 ## [1.26.6-beta.1](https://github.com/ScrapeGraphAI/Scrapegraph-ai/compare/v1.26.5...v1.26.6-beta.1) (2024-10-14)
@@ -79,7 +83,6 @@
 * remove variable "max_result" not being used in the code ([e76a68a](https://github.com/ScrapeGraphAI/Scrapegraph-ai/commit/e76a68a782e5bce48d421cb620d0b7bffa412918))
 
 * refactoring of gpt2 tokenizer ([44c3f9c](https://github.com/ScrapeGraphAI/Scrapegraph-ai/commit/44c3f9c98939c44caa86dc582242819a7c6a0f80))
->>>>>>> main
 
 ## [1.26.5](https://github.com/ScrapeGraphAI/Scrapegraph-ai/compare/v1.26.4...v1.26.5) (2024-10-13)
 

docs/source/introduction/overview.rst

@@ -22,6 +22,45 @@ This flexibility ensures that scrapers remain functional even when website layou
 We support many LLMs including **GPT, Gemini, Groq, Azure, Hugging Face** etc.
 as well as local models which can run on your machine using **Ollama**.
 
+AI Models and Token Limits
+==========================
+
+ScrapGraphAI supports a wide range of AI models from various providers. Each model has a specific token limit, which is important to consider when designing your scraping pipelines. Here's an overview of the supported models and their token limits:
+
+OpenAI Models
+-------------
+- GPT-3.5 Turbo (16,385 tokens)
+- GPT-4 (8,192 tokens)
+- GPT-4 Turbo Preview (128,000 tokens)
+
+Azure OpenAI Models
+-------------------
+- GPT-3.5 Turbo (16,385 tokens)
+- GPT-4 (8,192 tokens)
+- GPT-4 Turbo Preview (128,000 tokens)
+
+Google AI Models
+----------------
+- Gemini Pro (128,000 tokens)
+- Gemini 1.5 Pro (128,000 tokens)
+
+Anthropic Models
+----------------
+- Claude Instant (100,000 tokens)
+- Claude 2 (200,000 tokens)
+- Claude 3 (200,000 tokens)
+
+Mistral AI Models
+-----------------
+- Mistral Large (128,000 tokens)
+- Open Mistral 7B (32,000 tokens)
+- Open Mixtral 8x7B (32,000 tokens)
+
+For a complete list of supported models and their token limits, please refer to the API documentation.
+
+Understanding token limits is crucial for optimizing your scraping tasks. Larger token limits allow for processing more text in a single API call, which can be beneficial for scraping lengthy web pages or documents.
+
+
 Library Diagram
 ===============
 
@@ -95,4 +134,4 @@ Sponsors
 .. image:: ../../assets/transparent_stat.png
    :width: 15%
    :alt: Stat Proxies
-   :target: https://dashboard.statproxies.com/?refferal=scrapegraph
+   :target: https://dashboard.statproxies.com/?refferal=scrapegraph

docs/source/modules/modules.rst

@@ -5,3 +5,6 @@ scrapegraphai
    :maxdepth: 4
 
    scrapegraphai
+
+   scrapegraphai.helpers.models_tokens
+

docs/source/modules/scrapegraphai.helpers.models_tokens.rst

@@ -0,0 +1,28 @@
+scrapegraphai.helpers.models_tokens module
+==========================================
+
+.. automodule:: scrapegraphai.helpers.models_tokens
+   :members:
+   :undoc-members:
+   :show-inheritance:
+
+This module contains a comprehensive dictionary of AI models and their corresponding token limits. The `models_tokens` dictionary is organized by provider (e.g., OpenAI, Azure OpenAI, Google AI, etc.) and includes various models with their maximum token counts.
+
+Example usage:
+
+.. code-block:: python
+
+   from scrapegraphai.helpers.models_tokens import models_tokens
+
+   # Get the token limit for GPT-4
+   gpt4_limit = models_tokens['openai']['gpt-4']
+   print(f"GPT-4 token limit: {gpt4_limit}")
+
+   # Check the token limit for a specific model
+   model_name = "gpt-3.5-turbo"
+   if model_name in models_tokens['openai']:
+       print(f"{model_name} token limit: {models_tokens['openai'][model_name]}")
+   else:
+       print(f"{model_name} not found in the models list")
+
+This information is crucial for users to understand the capabilities and limitations of different AI models when designing their scraping pipelines.

pyproject.toml

@@ -4,6 +4,7 @@ name = "scrapegraphai"
 version = "1.27.0b7"
 
 
+
 description = "A web scraping library based on LangChain which uses LLM and direct graph logic to create scraping pipelines."
 authors = [
     { name = "Marco Vinciguerra", email = "[email protected]" },

scrapegraphai/utils/tokenizer.py

@@ -6,7 +6,6 @@
 from langchain_ollama import ChatOllama
 from langchain_mistralai import ChatMistralAI
 from langchain_core.language_models.chat_models import BaseChatModel
-from transformers import GPT2TokenizerFast
 
 def num_tokens_calculus(string: str, llm_model: BaseChatModel) -> int:
     """
@@ -24,13 +23,6 @@ def num_tokens_calculus(string: str, llm_model: BaseChatModel) -> int:
         from .tokenizers.tokenizer_ollama import num_tokens_ollama
         num_tokens_fn = num_tokens_ollama
 
-    elif isinstance(llm_model, GPT2TokenizerFast):
-        def num_tokens_gpt2(text: str, model: BaseChatModel) -> int:
-            tokenizer = GPT2TokenizerFast.from_pretrained("gpt2")
-            tokens = tokenizer.encode(text)
-            return len(tokens)
-        num_tokens_fn = num_tokens_gpt2
-
     else:
         from .tokenizers.tokenizer_openai import num_tokens_openai
         num_tokens_fn = num_tokens_openai

scrapegraphai/utils/tokenizers/tokenizer_ollama.py

@@ -3,7 +3,6 @@
 """
 from langchain_core.language_models.chat_models import BaseChatModel
 from ..logging import get_logger
-from transformers import GPT2TokenizerFast
 
 def num_tokens_ollama(text: str, llm_model:BaseChatModel) -> int:
     """
@@ -22,12 +21,8 @@ def num_tokens_ollama(text: str, llm_model:BaseChatModel) -> int:
 
     logger.debug(f"Counting tokens for text of {len(text)} characters")
 
-    if isinstance(llm_model, GPT2TokenizerFast):
-        tokenizer = GPT2TokenizerFast.from_pretrained("gpt2")
-        tokens = tokenizer.encode(text)
-        return len(tokens)
-
     # Use langchain token count implementation
     # NB: https://github.com/ollama/ollama/issues/1716#issuecomment-2074265507
     tokens = llm_model.get_num_tokens(text)
     return tokens
+