abstract syntax tree is what happens when CFGs is applied to text (i.e., the parser traces the AST in time).

Building an AST

Generally, each production will have an action, which is the relevant computation.

X -> Y1... Yn { action }

Each symbol X admits an attribute X.val

• for each terminal, val is the associated lexeme
• for non-terminals, val is the expression’s value (i.e., from the values of the subexpressions)

you job is to compose the subexpressions’ values and stick it into X.val. Notably, cycles are not allowed because we typically parse bottom-up, left ot right.

To determine

Question: can we find a good context permutation to improve reasoning capabilities.

One-Liner

Notable Methods

Two key evaluations:

• evalutanig relationships between gold documents; notice that performance relates to distance between documents (but FTing helps)
• investigate the effects between different attention masks (i.e., the use of prefix vs continuation masks)

IC Score

attention-based context attribution method

New Concepts

Key insight: correct answers will have single peak of IC scores at gold results; incorrect answers will have more dispersed IC scores.

Talks

• ACL2025 Keynote: Luke Zettemoyer
• ACL2025 Orals: Language Modeling 1
• ACL2025 Orals: QA

Posters

• ACL2025 Monday Morning Posters
• ACL2025 Tuesday Morning Posters
• ACL2025 Tuesday Afternoon Posters

Takes

• mayhaps we can apply thoughtbubbbles intutiton to BLT token pruning?