validmind 2.10.1__tar.gz → 2.11.0__tar.gz

{validmind-2.10.1 → validmind-2.11.0}/PKG-INFO

@@ -1,6 +1,6 @@
 Metadata-Version: 2.4
 Name: validmind
-Version: 2.10.1
+Version: 2.11.0
 Summary: ValidMind Library
 License: DUAL LICENSE NOTICE
          
@@ -732,12 +732,14 @@ Requires-Dist: mistune (>=3.0.2,<4.0.0)
 Requires-Dist: nest-asyncio (>=1.6.0,<2.0.0)
 Requires-Dist: nltk (>=3.8.1,<4.0.0) ; extra == "all"
 Requires-Dist: nltk (>=3.8.1,<4.0.0) ; extra == "nlp"
+Requires-Dist: numpy (>=1.23,<2.0.0)
 Requires-Dist: openai (>=1)
 Requires-Dist: pandas (>=2.0.3,<3.0.0)
-Requires-Dist: plotly (>=5.0.0)
+Requires-Dist: plotly (>=5.0.0,<6.0.0)
 Requires-Dist: polars
 Requires-Dist: presidio-analyzer ; extra == "pii-detection"
 Requires-Dist: presidio-structured ; extra == "pii-detection"
+Requires-Dist: pyarrow (<16) ; extra == "nlp"
 Requires-Dist: pycocoevalcap ; extra == "all"
 Requires-Dist: pycocoevalcap ; extra == "llm"
 Requires-Dist: python-dotenv
@@ -754,7 +756,6 @@ Requires-Dist: seaborn
 Requires-Dist: sentencepiece (>=0.2.0,<0.3.0) ; extra == "all"
 Requires-Dist: sentencepiece (>=0.2.0,<0.3.0) ; extra == "huggingface"
 Requires-Dist: sentencepiece (>=0.2.0,<0.3.0) ; extra == "llm"
-Requires-Dist: sentry-sdk (>=1.24.0,<2.0.0)
 Requires-Dist: shap (>=0.46.0) ; extra == "all"
 Requires-Dist: shap (>=0.46.0) ; extra == "explainability"
 Requires-Dist: statsmodels ; extra == "all"

{validmind-2.10.1 → validmind-2.11.0}/pyproject.toml

@@ -1,6 +1,6 @@
 [project]
 name = "validmind"
-version = "2.10.1"
+version = "2.11.0"
 description = "ValidMind Library"
 readme = "README.pypi.md"
 requires-python = ">=3.9,<3.13"
@@ -19,14 +19,14 @@ dependencies = [
   "matplotlib",
   "mistune (>=3.0.2,<4.0.0)",
   "nest-asyncio (>=1.6.0,<2.0.0)",
+  "numpy (>=1.23,<2.0.0)",
   "openai (>=1)",
   "pandas (>=2.0.3,<3.0.0)",
-  "plotly (>=5.0.0)",
+  "plotly (>=5.0.0,<6.0.0)",
   "polars",
   "python-dotenv",
   "scikit-learn",
   "seaborn",
-  "sentry-sdk (>=1.24.0,<2.0.0)",
   "tabulate (>=0.9.0,<0.10.0)",
   "tiktoken",
   "tqdm",
@@ -75,6 +75,7 @@ nlp = [
   "evaluate",
   "rouge (>=1)",
   "bert-score (>=0.3.13)",
+  "pyarrow (<16)",
 ]
 pytorch = ["torch (>=2.0.0)"]
 stats = ["scipy", "statsmodels", "arch"]

{validmind-2.10.1 → validmind-2.11.0}/validmind/__init__.py

@@ -21,7 +21,7 @@ To initialize the ValidMind Library, paste the code snippet with the model ident
 import validmind as vm
 
 vm.init(
-  api_host = "https://api.dev.vm.validmind.ai/api/v1/tracking/tracking",
+  api_host = "https://app.prod.validmind.ai/api/v1/tracking/tracking",
   api_key = "xxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxx",
   api_secret = "xxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxx",
   project = "<project-identifier>"

validmind-2.11.0/validmind/__version__.py

@@ -0,0 +1 @@
+__version__ = "2.11.0"

{validmind-2.10.1 → validmind-2.11.0}/validmind/ai/test_descriptions.py

@@ -5,9 +5,7 @@
 import json
 import os
 from concurrent.futures import ThreadPoolExecutor
-from typing import List, Optional, Union
-
-import tiktoken
+from typing import Any, Dict, List, Optional, Union
 
 from ..client_config import client_config
 from ..logging import get_logger
@@ -25,6 +23,21 @@ __executor = ThreadPoolExecutor()
 
 logger = get_logger(__name__)
 
+# Try to import tiktoken once at module load
+# Cache the result to avoid repeated import attempts
+_TIKTOKEN_AVAILABLE = False
+_TIKTOKEN_ENCODING = None
+
+try:
+    import tiktoken
+
+    _TIKTOKEN_ENCODING = tiktoken.encoding_for_model("gpt-4o")
+    _TIKTOKEN_AVAILABLE = True
+except (ImportError, Exception) as e:
+    logger.debug(
+        f"tiktoken unavailable, will use character-based token estimation: {e}"
+    )
+
 
 def _get_llm_global_context():
     # Get the context from the environment variable
@@ -42,6 +55,29 @@ def _get_llm_global_context():
     return context if context_enabled and context else None
 
 
+def _estimate_tokens_simple(text: str) -> int:
+    """Estimate token count using character-based heuristic.
+
+    Uses ~4 characters per token for English/JSON text.
+    This is a fallback when tiktoken is unavailable.
+    """
+    return len(text) // 4
+
+
+def _truncate_text_simple(text: str, max_tokens: int) -> str:
+    """Truncate text using character-based estimation."""
+    estimated_chars = max_tokens * 4
+    if len(text) <= estimated_chars:
+        return text
+
+    # Keep first portion and last 100 tokens worth (~400 chars)
+    # But ensure we don't take more than 25% for the tail
+    last_chars = min(400, estimated_chars // 4)
+    first_chars = estimated_chars - last_chars
+
+    return text[:first_chars] + "...[truncated]" + text[-last_chars:]
+
+
 def _truncate_summary(
     summary: Union[str, None], test_id: str, max_tokens: int = 100_000
 ):
@@ -49,20 +85,30 @@ def _truncate_summary(
         # since string itself is less than max_tokens, definitely small enough
         return summary
 
-    # TODO: better context length handling
-    encoding = tiktoken.encoding_for_model("gpt-4o")
-    summary_tokens = encoding.encode(summary)
+    if _TIKTOKEN_AVAILABLE:
+        # Use tiktoken for accurate token counting
+        summary_tokens = _TIKTOKEN_ENCODING.encode(summary)
 
-    if len(summary_tokens) > max_tokens:
-        logger.warning(
-            f"Truncating {test_id} due to context length restrictions..."
-            " Generated description may be innacurate"
-        )
-        summary = (
-            encoding.decode(summary_tokens[:max_tokens])
-            + "...[truncated]"
-            + encoding.decode(summary_tokens[-100:])
-        )
+        if len(summary_tokens) > max_tokens:
+            logger.warning(
+                f"Truncating {test_id} due to context length restrictions..."
+                " Generated description may be inaccurate"
+            )
+            summary = (
+                _TIKTOKEN_ENCODING.decode(summary_tokens[:max_tokens])
+                + "...[truncated]"
+                + _TIKTOKEN_ENCODING.decode(summary_tokens[-100:])
+            )
+    else:
+        # Fallback to character-based estimation
+        estimated_tokens = _estimate_tokens_simple(summary)
+
+        if estimated_tokens > max_tokens:
+            logger.warning(
+                f"Truncating {test_id} (estimated) due to context length restrictions..."
+                " Generated description may be inaccurate"
+            )
+            summary = _truncate_text_simple(summary, max_tokens)
 
     return summary
 
@@ -76,6 +122,7 @@ def generate_description(
     title: Optional[str] = None,
     instructions: Optional[str] = None,
     additional_context: Optional[str] = None,
+    params: Optional[Dict[str, Any]] = None,
 ):
     """Generate the description for the test results."""
     from validmind.api_client import generate_test_result_description
@@ -121,11 +168,10 @@ def generate_description(
             "test_description": test_description,
             "title": title,
             "summary": _truncate_summary(summary, test_id),
-            "figures": [
-                figure._get_b64_url() for figure in ([] if tables else figures)
-            ],
+            "figures": [figure._get_b64_url() for figure in figures or []],
             "additional_context": additional_context,
             "instructions": instructions,
+            "params": params,
         }
     )["content"]
 
@@ -139,6 +185,7 @@ def background_generate_description(
     title: Optional[str] = None,
     instructions: Optional[str] = None,
     additional_context: Optional[str] = None,
+    params: Optional[Dict[str, Any]] = None,
 ):
     def wrapped():
         try:
@@ -152,6 +199,7 @@ def background_generate_description(
                     title=title,
                     instructions=instructions,
                     additional_context=additional_context,
+                    params=params,
                 ),
                 True,
             )
@@ -183,6 +231,7 @@ def get_result_description(
     title: Optional[str] = None,
     instructions: Optional[str] = None,
     additional_context: Optional[str] = None,
+    params: Optional[Dict[str, Any]] = None,
 ):
     """Get the metadata dictionary for a test or metric result.
 
@@ -206,6 +255,7 @@ def get_result_description(
         should_generate (bool): Whether to generate the description or not. Defaults to True.
         instructions (Optional[str]): Instructions for the LLM to generate the description.
         additional_context (Optional[str]): Additional context for the LLM to generate the description.
+        params (Optional[Dict[str, Any]]): Test parameters used to customize test behavior.
 
     Returns:
         str: The description to be logged with the test results.
@@ -234,6 +284,7 @@ def get_result_description(
             title=title,
             instructions=_instructions,
             additional_context=additional_context,
+            params=params,
         )
 
     else:

{validmind-2.10.1 → validmind-2.11.0}/validmind/api_client.py

@@ -18,11 +18,10 @@ from urllib.parse import urlencode, urljoin
 import aiohttp
 import requests
 from aiohttp import FormData
-from ipywidgets import HTML, Accordion
 
 from .client_config import client_config
 from .errors import MissingAPICredentialsError, MissingModelIdError, raise_api_error
-from .logging import get_logger, init_sentry, log_api_operation, send_single_error
+from .logging import get_logger, log_api_operation
 from .utils import NumpyEncoder, is_html, md_to_html, run_async
 from .vm_models.figure import Figure
 
@@ -166,7 +165,7 @@ def _ping() -> Dict[str, Any]:
 
     client_info = r.json()
 
-    init_sentry(client_info.get("sentry_config", {}))
+    # Sentry removed: no telemetry initialization
 
     # Only show this confirmation the first time we connect to the API
     ack_connected = not client_config.model
@@ -244,14 +243,7 @@ def init(
 
 def reload():
     """Reconnect to the ValidMind API and reload the project configuration."""
-
-    try:
-        _ping()
-    except Exception as e:
-        # if the api host is https, assume we're not in dev mode and send to sentry
-        if _api_host.startswith("https://"):
-            send_single_error(e)
-        raise e
+    _ping()
 
 
 async def aget_metadata(content_id: str) -> Dict[str, Any]:
@@ -411,9 +403,7 @@ def log_input(input_id: str, type: str, metadata: Dict[str, Any]) -> Dict[str, A
     return run_async(alog_input, input_id, type, metadata)
 
 
-def log_text(
-    content_id: str, text: str, _json: Optional[Dict[str, Any]] = None
-) -> Dict[str, Any]:
+def log_text(content_id: str, text: str, _json: Optional[Dict[str, Any]] = None) -> str:
     """Logs free-form text to ValidMind API.
 
     Args:
@@ -426,7 +416,7 @@ def log_text(
         Exception: If the API call fails.
 
     Returns:
-        ipywidgets.Accordion: An accordion widget containing the logged text as HTML.
+        str: HTML string containing the logged text in an accordion format.
     """
     if not content_id or not isinstance(content_id, str):
         raise ValueError("`content_id` must be a non-empty string")
@@ -438,8 +428,10 @@ def log_text(
 
     log_text = run_async(alog_metadata, content_id, text, _json)
 
-    return Accordion(
-        children=[HTML(log_text["text"])],
+    from .vm_models.html_renderer import StatefulHTMLRenderer
+
+    return StatefulHTMLRenderer.render_accordion(
+        items=[log_text["text"]],
         titles=[f"Text Block: '{log_text['content_id']}'"],
     )