mikasenghaas

#!/usr/bin/env -S uv run --script
# /// script
# requires-python = ">=3.10"
# dependencies = ["renderers", "transformers>=4.46", "jinja2", "rich"]
# ///
"""
Replay the TITO failure modes from tito_failure_modes.py using
PrimeIntellect's `renderers` package (hand-coded per-model bridges)
instead of the naive `apply_chat_template`-delta algorithm.

mikasenghaas

#!/usr/bin/env -S uv run --script
# /// script
# requires-python = ">=3.10"
# dependencies = ["transformers", "jinja2", "rich"]
# ///
from transformers import AutoTokenizer
from rich.console import Console
from rich.text import Text

console = Console()

mikasenghaas

To reproduce, download the .json files with raw responses from here.

Running in standard mode

uv run oai_pydantic.py

Processing oai_response_1024.json...