kylelovesllms/grammar-v1-frames-ce

Grammar V1 — CE Populated Frames

The 40 depth-1 syntactic frames of grammar-v1-frames populated with randomly sampled lexicalizations, 100 base lexicalizations per (frame, task) (seeded RNG, seed=0; grammar source: experiments/01_grammar_v1/src/cfg/grammar.py).

Partition (README.md / DataGenFlow.png): ambiguous (in-distribution) frames contribute copy and translate examples split 90/10 into train/validation (per frame and task, so every ambiguous frame appears in both); unambiguous (out-of-distribution) frames contribute copy examples to train and translate examples to test. Copy and translate rows of a frame never share a sentence, so no test translate input appears as a train copy input.

Copy-task variant ce: copy simple (no RC) and center embedded (subject-attached RC) structures only

Train copy rows are kept only for frames whose relative clause attaches at one of these sites: None, Subject (None = no relative clause). The validation and test splits are identical across the ce-rb, ce and rb datasets (same base lexicalizations), so only the training data differs; validation therefore still contains copy rows of structures whose train copy rows were dropped. To keep the train token budget equal to ce-rb's (non-pad tokens the model is trained on: <bos> src <task> <sep> tgt <eos> = 2 * n_tokens + 4 per row), every remaining train (frame, task) group is boosted by the same factor with additional distinct lexicalizations (rounded per group):

Dataset	Train rows	Train tokens	Boost factor
ce-rb (baseline)	4,960	78,080	1
ce	5,112	78,056	1.3315

• train: 5112 rows = 1440 translate (ambiguous) + 480 copy (ambiguous) + 3192 copy (unambiguous)
• validation: 240 rows = 120 translate + 120 copy (ambiguous only)
• test: 2800 rows (translate, unambiguous)

Train rows per structure

Task	Attachment Site	Gap Type	Frames	Rows per frame	Rows	Tokens
translate	None	None	4	120	480	4,800
translate	Object	Subject Gap	8	120	960	16,320
copy	None	None	4	120	480	4,800
copy	Subject	Subject Gap	16	133	2,128	34,048
copy	Subject	Object Gap	8	133	1,064	18,088
Total					5,112	78,056

HF Sentence is the task target: the HI sentence itself for copy rows, the hierarchical (Head Final) translation for translate rows. Frames use indexed bracketed notation where [N_3] means the terminal filled by HI token 3.

Frame counts per category (before population)

Attachment Site	Gap Type	Frames	Ambiguous	Unambiguous
None	None	4	4	0
Subject	Subject Gap	16	0	16
Subject	Object Gap	8	0	8
Object	Subject Gap	8	8	0
Object	Object Gap	4	0	4
Total		40	12	28