Skip to content

fix(memory): count tokens for ToolCallBlock, ThinkingBlock, CitableBlock, CitationBlock #22153

New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

Sign up for GitHub

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Open
citizen204 wants to merge 1 commit into
base: main
Choose a base branch
from
+32 −2
Open

fix(memory): count tokens for ToolCallBlock, ThinkingBlock, CitableBlock, CitationBlock #22153

Changes from all commits
Commits
Show all changes
1 commit
Select commit
File filter

Filter by extension

Filter by extension
Conversations
Failed to load comments.
Loading
Jump to
Jump to file
Failed to load files.
Loading
Diff view
Diff view
34 changes: 32 additions & 2 deletions llama-index-core/llama_index/core/memory/memory.py
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -358,11 +358,12 @@ def _estimate_token_count(
CitableBlock,
CitationBlock,
ThinkingBlock,
ToolCallBlock,
]
] = []

for block in message_or_blocks.blocks:
if not isinstance(block, (CachePoint, ToolCallBlock)):
if not isinstance(block, CachePoint):
blocks.append(block)

# Estimate the token count for the additional kwargs
Expand All @@ -380,7 +381,7 @@ def _estimate_token_count(
blocks = []
for msg in messages:
for block in msg.blocks:
if not isinstance(block, (CachePoint, ToolCallBlock)):
if not isinstance(block, CachePoint):
blocks.append(block)

# Estimate the token count for the additional kwargs
Expand Down Expand Up @@ -437,6 +438,35 @@ def _estimate_token_count(
token_count += self.audio_token_size_estimate
elif isinstance(block, DocumentBlock):
token_count += self.document_token_size_estimate
elif isinstance(block, ToolCallBlock):
token_count += len(
self.tokenizer_fn(
block.tool_name + " " + str(block.tool_kwargs)
)
)
elif isinstance(block, ThinkingBlock):
token_count += block.num_tokens or len(
self.tokenizer_fn(block.content or "")
)
elif isinstance(block, CitableBlock):
token_count += len(
self.tokenizer_fn(block.title + " " + block.source)
)
for inner in block.content:
if isinstance(inner, TextBlock):
token_count += len(self.tokenizer_fn(inner.text))
elif isinstance(inner, ImageBlock):
token_count += self.image_token_size_estimate
else:
token_count += self.document_token_size_estimate
elif isinstance(block, CitationBlock):
token_count += len(
self.tokenizer_fn(block.title + " " + block.source)
)
if isinstance(block.cited_content, TextBlock):
token_count += len(self.tokenizer_fn(block.cited_content.text))
else:
token_count += self.image_token_size_estimate

return token_count

Expand Down
Loading