Token counter telemetry
Understand token-count precision in each result.
Inspect the counter
token_counter_used reports the process-level active counter: tiktoken or heuristic.
tokens.py
from llmslim import compress
result = compress(long_text)
print(result.token_counter_used)When is the heuristic used?
When the process cannot load the tiktoken encoder. LLMSlim emits one warning per process.