full evaluation artifacts: entry + ablation families, per-item predictions, analysis tooling

README.md

@@ -0,0 +1,84 @@
+---
+license: cc-by-4.0
+language: en
+tags:
+  - babylm
+  - babylm-2026
+  - strict-small
+  - evaluation-artifacts
+  - reproducibility
+pretty_name: bind1 (BabyLM 2026 Strict-Small) — full evaluation artifacts
+---
+
+# bind1 — BabyLM 2026 Strict-Small entry: full evaluation artifacts
+
+Complete, per-item evaluation outputs for the
+[`SecludedCorner/bind1-babylm2026-strict-small`](https://huggingface.co/SecludedCorner/bind1-babylm2026-strict-small)
+entry and its ablation families, released so that **every number we report is downloadable and
+recomputable** — including the negative results.
+
+All outputs were produced by the official `babylm-eval` pipeline (zero-shot backend `causal`;
+GLUE via the official fine-tuning protocol). Reports are the pipeline's own
+`best_temperature_report.txt`; predictions are per-item.
+
+## Model families included (35 evaluation targets)
+
+| Directory pattern | What it is | What it shows |
+|---|---|---|
+| `hf_loop2_full(_sN)` | the entry architecture, trained loop depth T=3, seeds 0–3 | main results |
+| `hf_loop2_T1/T2/T4(_sN)` | same weights, forced inference depth T=1/2/4 | the T-sweep: iteration helps grammar in all seeds, erodes healthy-write entity tracking, and partially rebuilds the one collapsed write |
+| `hf_loop2_sever(_sN)` | same weights, verdict-to-trust edge zeroed at inference | inference-inertness of the feedback edge (4/4 seeds) |
+| `hf_loop2_novg(_sN)` (+`_ti1`) | retrained from scratch with the edge frozen | training-scaffold evidence (seed-conditional: 3/4) |
+| `hf_baseline_full(_sN)` | size-matched plain-decoder monolith, seeds 0–3 | the paired control for every claim |
+| `hf_loop_full` | the earlier bypassed variant (v=1.0, loop inert) | the audit-caught failure kept as a diagnostic |
+
+Internal (logical) experiment ids map to these physical names as
+`bind1_tt3_sN` = `hf_loop2_full(_sN)`, `bind1_tt3_sN_ti{K}` = `hf_loop2_T{K}(_sN)`,
+`mono_sN` = `hf_baseline_full(_sN)`, `bind1_tt3_sN_novg` = `hf_loop2_novg(_sN)`.
+
+## Layout
+
+```
+results/<model>/zero_shot/causal/<task>/<subset>/
+    best_temperature_report.txt   # official aggregate report
+    predictions.json              # per-item predictions
+results/hf_loop2_full/finetune/<task>/results.txt      # official GLUE protocol (entry)
+results/hf_baseline_full/finetune/<task>/results.txt   # official GLUE protocol (control)
+scripts/oracle_best_of_T.py       # best-of-T oracle + chance-null recovery analysis
+scripts/make_t_variant.py         # builds the forced-T inference variants (config-only clones)
+patches/compute_results_per_item_is_correct.diff  # additive patch to the official pipeline
+```
+
+## Per-item correctness (`is_correct`)
+
+Entity-tracking `predictions.json` records include `pred_idx` / `gold_idx` / `is_correct`,
+added by the **additive** patch in `patches/` (the official `id`/`pred` fields and all scoring
+logic are untouched; patched reruns reproduce the official aggregates exactly, 12/12 checked).
+This is what makes the chance-null analysis possible: for answers the single-pass model got
+wrong and iteration *changed*, a memoryless re-roll lands on gold ~25% of the time (5 options).
+Recovery at ~25% = re-rolled noise; recovery far above = real signal. Both verdicts occur in
+this data — ~25% in the three healthy-write seeds, 37–45% in the collapsed-write seed —
+which is the evidence behind the model card's "iteration as a backup pathway" note.
+
+Recompute it yourself:
+
+```bash
+python scripts/oracle_best_of_T.py bind1_tt3_s0   # any of s0..s3
+```
+
+(The script expects the experiment-archive layout; adapt the paths at the top, or read the
+predictions files directly — the schema is one JSON object per subset with a `predictions` list.)
+
+## Scope note (honest boundary)
+
+These are the artifacts of the **2026 entry generation** (bind1). Artifacts of the
+training-time-depth study that motivates the 2027 generation are withheld until a companion
+paper clears double-blind review, and will be added here afterwards. Nothing in this dataset
+was selected post-hoc: every ablation family we ran on the entry generation is present,
+including the ones that falsified our own pre-registered predictions.
+
+## Citation
+
+Yulin Yang (ORCID 0009-0007-4827-8449). *A Microkernel Language Model: Reasoning as
+Mutually-Supporting Aggregates, and Why Its Parts Must Be Judged Together.* BabyLM Challenge
+2026 (Strict-Small track) submission.

patches/compute_results_per_item_is_correct.diff

@@ -0,0 +1,26 @@
+diff --git a/strict/evaluation_pipeline/sentence_zero_shot/compute_results.py b/strict/evaluation_pipeline/sentence_zero_shot/compute_results.py
+index c407590..2b91b48 100644
+--- a/strict/evaluation_pipeline/sentence_zero_shot/compute_results.py
++++ b/strict/evaluation_pipeline/sentence_zero_shot/compute_results.py
+@@ -78,9 +78,19 @@ def rank_and_evaluate(args, subset_to_stats, all_log_probs, raw_sentences, label
+             if args.save_predictions:
+                 num_id_matches = len(predictions[temp][uid])
+                 if args.task in ("comps", "ewok"):
+-                    predictions[temp][uid].append({"id" : f"{uid}_{num_id_matches}", "pred" : raw_sentence_dict["sentences"][chosen_sentence]})
++                    pred_text = raw_sentence_dict["sentences"][chosen_sentence]
+                 else:
+-                    predictions[temp][uid].append({"id" : f"{uid}_{num_id_matches}", "pred" : raw_sentence_dict["completions"][chosen_sentence]})
++                    pred_text = raw_sentence_dict["completions"][chosen_sentence]
++                # pred_idx/gold_idx/is_correct are additive fields (id/pred schema unchanged):
++                # the ranked choice is decided here and cannot be reliably reconstructed from
++                # pred text alone (duplicate options, per-UID counter != example_id).
++                predictions[temp][uid].append({
++                    "id" : f"{uid}_{num_id_matches}",
++                    "pred" : pred_text,
++                    "pred_idx" : int(chosen_sentence),
++                    "gold_idx" : int(label),
++                    "is_correct" : bool(is_correct),
++                })
+ 
+ def compute_causal_results(args, model, dataloader, temperatures):
+     subset_to_stats = {temp : {} for temp in temperatures}

results/hf_baseline_full/finetune/boolq/results.txt

@@ -0,0 +1,3 @@
+accuracy: 0.673394495412844
+f1: 0.7532347504621072
+mcc: 0.2731513822415415

results/hf_baseline_full/finetune/mnli/results.txt

@@ -0,0 +1 @@
+accuracy: 0.43989405052974734

results/hf_baseline_full/finetune/mrpc/results.txt

@@ -0,0 +1,3 @@
+accuracy: 0.6862745098039216
+f1: 0.8095238095238095
+mcc: 0.10227299881383177

results/hf_baseline_full/finetune/multirc/results.txt

@@ -0,0 +1,3 @@
+accuracy: 0.5742574257425742
+f1: 0.07194244604316546
+mcc: 0.021875612870243824

results/hf_baseline_full/finetune/qqp/results.txt

@@ -0,0 +1,3 @@
+accuracy: 0.7104625278258718
+f1: 0.600668622501194
+mcc: 0.37415736608969374

results/hf_baseline_full/finetune/rte/results.txt

@@ -0,0 +1,3 @@
+accuracy: 0.5107913669064749
+f1: 0.48484848484848486
+mcc: 0.019941920640703076

results/hf_baseline_full/finetune/wsc/results.txt

@@ -0,0 +1,3 @@
+accuracy: 0.6923076923076923
+f1: 0.5294117647058824
+mcc: 0.3221899920504105

results/hf_baseline_full/zero_shot/causal/blimp/blimp_filtered/best_temperature_report.txt

@@ -0,0 +1,95 @@
+TEMPERATURE: 1.00
+
+### FIELD ACCURACY
+syntax: 62.32
+morphology: 70.77
+syntax/semantics: 63.44
+semantics: 58.39
+
+### UID ACCURACY
+adjunct_island: 79.85
+anaphor_gender_agreement: 70.65
+anaphor_number_agreement: 81.31
+animate_subject_passive: 69.72
+animate_subject_trans: 82.56
+causative: 49.63
+complex_NP_island: 53.66
+coordinate_structure_constraint_complex_left_branch: 31.35
+coordinate_structure_constraint_object_extraction: 54.27
+determiner_noun_agreement_1: 83.85
+determiner_noun_agreement_2: 84.96
+determiner_noun_agreement_irregular_1: 71.81
+determiner_noun_agreement_irregular_2: 83.54
+determiner_noun_agreement_with_adjective_1: 78.14
+determiner_noun_agreement_with_adj_2: 84.38
+determiner_noun_agreement_with_adj_irregular_1: 77.72
+determiner_noun_agreement_with_adj_irregular_2: 78.45
+distractor_agreement_relational_noun: 47.08
+distractor_agreement_relative_clause: 40.07
+drop_argument: 71.09
+ellipsis_n_bar_1: 67.46
+ellipsis_n_bar_2: 69.57
+existential_there_object_raising: 65.02
+existential_there_quantifiers_1: 92.37
+existential_there_quantifiers_2: 45.66
+existential_there_subject_raising: 69.70
+expletive_it_object_raising: 68.12
+inchoative: 42.81
+intransitive: 64.06
+irregular_past_participle_adjectives: 51.72
+irregular_past_participle_verbs: 72.93
+irregular_plural_subject_verb_agreement_1: 66.17
+irregular_plural_subject_verb_agreement_2: 68.83
+left_branch_island_echo_question: 49.63
+left_branch_island_simple_question: 44.69
+matrix_question_npi_licensor_present: 11.30
+npi_present_1: 31.90
+npi_present_2: 44.97
+only_npi_licensor_present: 45.01
+only_npi_scope: 38.71
+passive_1: 79.05
+passive_2: 73.86
+principle_A_case_1: 100.00
+principle_A_case_2: 72.68
+principle_A_c_command: 59.41
+principle_A_domain_1: 93.76
+principle_A_domain_2: 61.53
+principle_A_domain_3: 52.71
+principle_A_reconstruction: 37.33
+regular_plural_subject_verb_agreement_1: 68.76
+regular_plural_subject_verb_agreement_2: 61.80
+sentential_negation_npi_licensor_present: 98.69
+sentential_negation_npi_scope: 52.24
+sentential_subject_island: 44.12
+superlative_quantifiers_1: 72.83
+superlative_quantifiers_2: 79.41
+tough_vs_raising_1: 36.92
+tough_vs_raising_2: 81.30
+transitive: 64.40
+wh_island: 57.50
+wh_questions_object_gap: 67.40
+wh_questions_subject_gap: 87.08
+wh_questions_subject_gap_long_distance: 88.68
+wh_vs_that_no_gap: 92.10
+wh_vs_that_no_gap_long_distance: 97.26
+wh_vs_that_with_gap: 34.82
+wh_vs_that_with_gap_long_distance: 10.99
+
+### LINGUISTICS_TERM ACCURACY
+island_effects: 51.89
+anaphor_agreement: 75.87
+s-selection: 76.24
+argument_structure: 63.77
+determiner_noun_agreement: 80.73
+subject_verb_agreement: 59.00
+ellipsis: 68.53
+control_raising: 63.88
+quantifiers: 72.81
+irregular_forms: 62.22
+npi_licensing: 46.14
+binding: 67.83
+filler_gap_dependency: 67.73
+
+### AVERAGE ACCURACY
+64.35
+

results/hf_baseline_full/zero_shot/causal/blimp/supplement_filtered/best_temperature_report.txt

@@ -0,0 +1,18 @@
+TEMPERATURE: 1.00
+
+### FIELD ACCURACY
+supplement: 73.28
+
+### UID ACCURACY
+hypernym: 49.05
+qa_congruence_easy: 56.25
+qa_congruence_tricky: 45.45
+subject_aux_inversion: 80.92
+turn_taking: 61.07
+
+### LINGUISTICS_TERM ACCURACY
+supplement: 73.28
+
+### AVERAGE ACCURACY
+58.55
+

results/hf_baseline_full/zero_shot/causal/comps/comps/best_temperature_report.txt

@@ -0,0 +1,11 @@
+TEMPERATURE: 1.00
+
+### UID ACCURACY
+base: 52.65
+wugs: 51.40
+wugs_dist_before: 53.40
+wugs_dist_in_between: 46.53
+
+### AVERAGE ACCURACY
+51.00
+

results/hf_baseline_full/zero_shot/causal/entity_tracking/entity_tracking/best_temperature_report.txt

@@ -0,0 +1,28 @@
+TEMPERATURE: 1.00
+
+### UID ACCURACY
+ambiref_0_ops: 30.64
+ambiref_1_ops: 27.68
+ambiref_2_ops: 26.49
+ambiref_3_ops: 28.81
+ambiref_4_ops: 25.85
+ambiref_5_ops: 32.62
+move_contents_0_ops: 28.60
+move_contents_1_ops: 24.09
+move_contents_2_ops: 30.41
+move_contents_3_ops: 27.72
+move_contents_4_ops: 28.92
+move_contents_5_ops: 26.28
+regular_0_ops: 30.20
+regular_1_ops: 23.06
+regular_2_ops: 26.90
+regular_3_ops: 24.92
+regular_4_ops: 26.76
+regular_5_ops: 30.82
+regular: 27.11
+ambiref: 28.68
+move_contents: 27.67
+
+### AVERAGE ACCURACY
+27.82
+

results/hf_baseline_full_s1/zero_shot/causal/blimp/blimp_filtered/best_temperature_report.txt

@@ -0,0 +1,95 @@
+TEMPERATURE: 1.00
+
+### FIELD ACCURACY
+syntax: 63.32
+morphology: 72.97
+syntax/semantics: 65.49
+semantics: 63.25
+
+### UID ACCURACY
+adjunct_island: 74.57
+anaphor_gender_agreement: 65.60
+anaphor_number_agreement: 83.89
+animate_subject_passive: 66.03
+animate_subject_trans: 82.67
+causative: 48.29
+complex_NP_island: 48.35
+coordinate_structure_constraint_complex_left_branch: 47.57
+coordinate_structure_constraint_object_extraction: 53.74
+determiner_noun_agreement_1: 85.04
+determiner_noun_agreement_2: 81.85
+determiner_noun_agreement_irregular_1: 74.01
+determiner_noun_agreement_irregular_2: 80.61
+determiner_noun_agreement_with_adjective_1: 79.42
+determiner_noun_agreement_with_adj_2: 77.05
+determiner_noun_agreement_with_adj_irregular_1: 73.54
+determiner_noun_agreement_with_adj_irregular_2: 75.71
+distractor_agreement_relational_noun: 42.77
+distractor_agreement_relative_clause: 33.18
+drop_argument: 70.00
+ellipsis_n_bar_1: 67.83
+ellipsis_n_bar_2: 69.32
+existential_there_object_raising: 70.57
+existential_there_quantifiers_1: 89.78
+existential_there_quantifiers_2: 58.40
+existential_there_subject_raising: 73.70
+expletive_it_object_raising: 70.36
+inchoative: 42.34
+intransitive: 60.71
+irregular_past_participle_adjectives: 82.00
+irregular_past_participle_verbs: 77.71
+irregular_plural_subject_verb_agreement_1: 73.01
+irregular_plural_subject_verb_agreement_2: 79.48
+left_branch_island_echo_question: 59.03
+left_branch_island_simple_question: 61.20
+matrix_question_npi_licensor_present: 17.44
+npi_present_1: 47.85
+npi_present_2: 48.03
+only_npi_licensor_present: 61.22
+only_npi_scope: 66.67
+passive_1: 77.62
+passive_2: 76.41
+principle_A_case_1: 99.89
+principle_A_case_2: 73.88
+principle_A_c_command: 50.42
+principle_A_domain_1: 97.48
+principle_A_domain_2: 67.87
+principle_A_domain_3: 54.84
+principle_A_reconstruction: 46.23
+regular_plural_subject_verb_agreement_1: 72.13
+regular_plural_subject_verb_agreement_2: 71.11
+sentential_negation_npi_licensor_present: 95.65
+sentential_negation_npi_scope: 34.56
+sentential_subject_island: 32.88
+superlative_quantifiers_1: 60.27
+superlative_quantifiers_2: 88.74
+tough_vs_raising_1: 40.72
+tough_vs_raising_2: 72.50
+transitive: 65.09
+wh_island: 69.06
+wh_questions_object_gap: 74.97
+wh_questions_subject_gap: 91.09
+wh_questions_subject_gap_long_distance: 91.72
+wh_vs_that_no_gap: 93.15
+wh_vs_that_no_gap_long_distance: 97.14
+wh_vs_that_with_gap: 25.57
+wh_vs_that_with_gap_long_distance: 5.71
+
+### LINGUISTICS_TERM ACCURACY
+island_effects: 55.88
+anaphor_agreement: 74.55
+s-selection: 74.48
+argument_structure: 63.16
+determiner_noun_agreement: 78.71
+subject_verb_agreement: 62.35
+ellipsis: 68.59
+control_raising: 65.12
+quantifiers: 74.41
+irregular_forms: 79.87
+npi_licensing: 52.93
+binding: 69.72
+filler_gap_dependency: 67.76
+
+### AVERAGE ACCURACY
+66.38
+

results/hf_baseline_full_s1/zero_shot/causal/blimp/supplement_filtered/best_temperature_report.txt

@@ -0,0 +1,18 @@
+TEMPERATURE: 1.00
+
+### FIELD ACCURACY
+supplement: 77.10
+
+### UID ACCURACY
+hypernym: 49.17
+qa_congruence_easy: 59.38
+qa_congruence_tricky: 43.64
+subject_aux_inversion: 86.11
+turn_taking: 60.36
+
+### LINGUISTICS_TERM ACCURACY
+supplement: 77.10
+
+### AVERAGE ACCURACY
+59.73
+

results/hf_baseline_full_s1/zero_shot/causal/entity_tracking/entity_tracking/best_temperature_report.txt

@@ -0,0 +1,28 @@
+TEMPERATURE: 1.00
+
+### UID ACCURACY
+ambiref_0_ops: 25.04
+ambiref_1_ops: 25.86
+ambiref_2_ops: 26.82
+ambiref_3_ops: 27.32
+ambiref_4_ops: 22.11
+ambiref_5_ops: 32.62
+move_contents_0_ops: 21.32
+move_contents_1_ops: 16.34
+move_contents_2_ops: 15.37
+move_contents_3_ops: 15.02
+move_contents_4_ops: 13.61
+move_contents_5_ops: 19.23
+regular_0_ops: 17.33
+regular_1_ops: 14.33
+regular_2_ops: 16.34
+regular_3_ops: 19.80
+regular_4_ops: 20.25
+regular_5_ops: 22.01
+regular: 18.34
+ambiref: 26.63
+move_contents: 16.81
+
+### AVERAGE ACCURACY
+20.60
+

results/hf_baseline_full_s2/zero_shot/causal/blimp/blimp_filtered/best_temperature_report.txt

@@ -0,0 +1,95 @@
+TEMPERATURE: 1.00
+
+### FIELD ACCURACY
+syntax: 62.02
+morphology: 73.07
+syntax/semantics: 63.41
+semantics: 55.49
+
+### UID ACCURACY
+adjunct_island: 75.97
+anaphor_gender_agreement: 75.08
+anaphor_number_agreement: 85.07
+animate_subject_passive: 65.47
+animate_subject_trans: 84.18
+causative: 49.39
+complex_NP_island: 46.22
+coordinate_structure_constraint_complex_left_branch: 36.31
+coordinate_structure_constraint_object_extraction: 61.43
+determiner_noun_agreement_1: 87.51
+determiner_noun_agreement_2: 86.04
+determiner_noun_agreement_irregular_1: 76.36
+determiner_noun_agreement_irregular_2: 84.88
+determiner_noun_agreement_with_adjective_1: 81.99
+determiner_noun_agreement_with_adj_2: 76.73
+determiner_noun_agreement_with_adj_irregular_1: 74.79
+determiner_noun_agreement_with_adj_irregular_2: 76.31
+distractor_agreement_relational_noun: 48.10
+distractor_agreement_relative_clause: 38.12
+drop_argument: 71.30
+ellipsis_n_bar_1: 66.08
+ellipsis_n_bar_2: 71.26
+existential_there_object_raising: 64.90
+existential_there_quantifiers_1: 88.60
+existential_there_quantifiers_2: 52.80
+existential_there_subject_raising: 75.97
+expletive_it_object_raising: 67.59
+inchoative: 47.02
+intransitive: 63.36
+irregular_past_participle_adjectives: 57.34
+irregular_past_participle_verbs: 79.19
+irregular_plural_subject_verb_agreement_1: 67.79
+irregular_plural_subject_verb_agreement_2: 75.56
+left_branch_island_echo_question: 43.08
+left_branch_island_simple_question: 43.85
+matrix_question_npi_licensor_present: 13.78
+npi_present_1: 39.49
+npi_present_2: 44.86
+only_npi_licensor_present: 61.11
+only_npi_scope: 55.32
+passive_1: 80.00
+passive_2: 76.63
+principle_A_case_1: 99.89
+principle_A_case_2: 67.87
+principle_A_c_command: 52.11
+principle_A_domain_1: 95.73
+principle_A_domain_2: 59.02
+principle_A_domain_3: 50.58
+principle_A_reconstruction: 44.78
+regular_plural_subject_verb_agreement_1: 71.80
+regular_plural_subject_verb_agreement_2: 69.31
+sentential_negation_npi_licensor_present: 99.78
+sentential_negation_npi_scope: 41.79
+sentential_subject_island: 43.70
+superlative_quantifiers_1: 39.43
+superlative_quantifiers_2: 60.24
+tough_vs_raising_1: 38.08
+tough_vs_raising_2: 75.98
+transitive: 65.55
+wh_island: 55.52
+wh_questions_object_gap: 74.04
+wh_questions_subject_gap: 82.29
+wh_questions_subject_gap_long_distance: 90.90
+wh_vs_that_no_gap: 93.84
+wh_vs_that_no_gap_long_distance: 93.71
+wh_vs_that_with_gap: 31.66
+wh_vs_that_with_gap_long_distance: 7.47
+
+### LINGUISTICS_TERM ACCURACY
+island_effects: 50.83
+anaphor_agreement: 79.97
+s-selection: 74.97
+argument_structure: 64.97
+determiner_noun_agreement: 80.89
+subject_verb_agreement: 62.12
+ellipsis: 68.71
+control_raising: 64.22
+quantifiers: 60.04
+irregular_forms: 68.16
+npi_licensing: 50.79
+binding: 66.80
+filler_gap_dependency: 67.02
+
+### AVERAGE ACCURACY
+64.51
+

results/hf_baseline_full_s2/zero_shot/causal/blimp/supplement_filtered/best_temperature_report.txt

@@ -0,0 +1,18 @@
+TEMPERATURE: 1.00
+
+### FIELD ACCURACY
+supplement: 75.28
+
+### UID ACCURACY
+hypernym: 52.97
+qa_congruence_easy: 59.38
+qa_congruence_tricky: 46.06
+subject_aux_inversion: 82.73
+turn_taking: 60.36
+
+### LINGUISTICS_TERM ACCURACY
+supplement: 75.28
+
+### AVERAGE ACCURACY
+60.30
+

results/hf_baseline_full_s2/zero_shot/causal/entity_tracking/entity_tracking/best_temperature_report.txt

@@ -0,0 +1,28 @@
+TEMPERATURE: 1.00
+
+### UID ACCURACY
+ambiref_0_ops: 25.37
+ambiref_1_ops: 22.41
+ambiref_2_ops: 26.32
+ambiref_3_ops: 28.81
+ambiref_4_ops: 31.71
+ambiref_5_ops: 35.83
+move_contents_0_ops: 23.64
+move_contents_1_ops: 14.52
+move_contents_2_ops: 26.78
+move_contents_3_ops: 23.43
+move_contents_4_ops: 24.95
+move_contents_5_ops: 30.13
+regular_0_ops: 24.75
+regular_1_ops: 19.60
+regular_2_ops: 25.41
+regular_3_ops: 25.25
+regular_4_ops: 26.23
+regular_5_ops: 31.45
+regular: 25.45
+ambiref: 28.41
+move_contents: 23.91
+
+### AVERAGE ACCURACY
+25.92
+

results/hf_baseline_full_s3/zero_shot/causal/blimp/blimp_filtered/best_temperature_report.txt

@@ -0,0 +1,95 @@
+TEMPERATURE: 1.00
+
+### FIELD ACCURACY
+syntax: 61.49
+morphology: 71.64
+syntax/semantics: 64.36
+semantics: 64.11
+
+### UID ACCURACY
+adjunct_island: 73.60
+anaphor_gender_agreement: 76.62
+anaphor_number_agreement: 80.13
+animate_subject_passive: 65.81
+animate_subject_trans: 85.37
+causative: 50.12
+complex_NP_island: 44.80
+coordinate_structure_constraint_complex_left_branch: 38.74
+coordinate_structure_constraint_object_extraction: 57.64
+determiner_noun_agreement_1: 83.96
+determiner_noun_agreement_2: 83.57
+determiner_noun_agreement_irregular_1: 76.06
+determiner_noun_agreement_irregular_2: 80.00
+determiner_noun_agreement_with_adjective_1: 80.39
+determiner_noun_agreement_with_adj_2: 76.51
+determiner_noun_agreement_with_adj_irregular_1: 73.68
+determiner_noun_agreement_with_adj_irregular_2: 75.24
+distractor_agreement_relational_noun: 54.70
+distractor_agreement_relative_clause: 41.91
+drop_argument: 70.00
+ellipsis_n_bar_1: 66.58
+ellipsis_n_bar_2: 72.46
+existential_there_object_raising: 69.95
+existential_there_quantifiers_1: 89.57
+existential_there_quantifiers_2: 37.76
+existential_there_subject_raising: 74.78
+expletive_it_object_raising: 70.09
+inchoative: 45.73
+intransitive: 62.79
+irregular_past_participle_adjectives: 62.02
+irregular_past_participle_verbs: 77.07
+irregular_plural_subject_verb_agreement_1: 65.17
+irregular_plural_subject_verb_agreement_2: 69.39
+left_branch_island_echo_question: 51.21
+left_branch_island_simple_question: 43.43
+matrix_question_npi_licensor_present: 19.16
+npi_present_1: 40.37
+npi_present_2: 54.27
+only_npi_licensor_present: 62.47
+only_npi_scope: 60.22
+passive_1: 77.02
+passive_2: 73.86
+principle_A_case_1: 100.00
+principle_A_case_2: 75.30
+principle_A_c_command: 56.45
+principle_A_domain_1: 93.22
+principle_A_domain_2: 61.97
+principle_A_domain_3: 52.28
+principle_A_reconstruction: 34.64
+regular_plural_subject_verb_agreement_1: 62.47
+regular_plural_subject_verb_agreement_2: 67.72
+sentential_negation_npi_licensor_present: 99.46
+sentential_negation_npi_scope: 43.86
+sentential_subject_island: 30.70
+superlative_quantifiers_1: 85.29
+superlative_quantifiers_2: 85.29
+tough_vs_raising_1: 36.71
+tough_vs_raising_2: 75.11
+transitive: 65.21
+wh_island: 65.00
+wh_questions_object_gap: 69.27
+wh_questions_subject_gap: 81.18
+wh_questions_subject_gap_long_distance: 85.06
+wh_vs_that_no_gap: 91.17
+wh_vs_that_no_gap_long_distance: 94.97
+wh_vs_that_with_gap: 32.86
+wh_vs_that_with_gap_long_distance: 11.32
+
+### LINGUISTICS_TERM ACCURACY
+island_effects: 50.71
+anaphor_agreement: 78.34
+s-selection: 75.74
+argument_structure: 63.74
+determiner_noun_agreement: 78.95
+subject_verb_agreement: 60.40
+ellipsis: 69.57
+control_raising: 64.86
+quantifiers: 74.96
+irregular_forms: 69.47
+npi_licensing: 54.18
+binding: 67.30
+filler_gap_dependency: 65.93
+
+### AVERAGE ACCURACY
+65.24
+

results/hf_baseline_full_s3/zero_shot/causal/blimp/supplement_filtered/best_temperature_report.txt

@@ -0,0 +1,18 @@
+TEMPERATURE: 1.00
+
+### FIELD ACCURACY
+supplement: 75.58
+
+### UID ACCURACY
+hypernym: 51.07
+qa_congruence_easy: 54.69
+qa_congruence_tricky: 44.24
+subject_aux_inversion: 83.71
+turn_taking: 60.36
+
+### LINGUISTICS_TERM ACCURACY
+supplement: 75.58
+
+### AVERAGE ACCURACY
+58.81
+

results/hf_baseline_full_s3/zero_shot/causal/entity_tracking/entity_tracking/best_temperature_report.txt

@@ -0,0 +1,28 @@
+TEMPERATURE: 1.00
+
+### UID ACCURACY
+ambiref_0_ops: 16.64
+ambiref_1_ops: 12.85
+ambiref_2_ops: 14.90
+ambiref_3_ops: 13.41
+ambiref_4_ops: 17.24
+ambiref_5_ops: 11.23
+move_contents_0_ops: 19.83
+move_contents_1_ops: 13.53
+move_contents_2_ops: 11.24
+move_contents_3_ops: 13.04
+move_contents_4_ops: 13.04
+move_contents_5_ops: 18.59
+regular_0_ops: 16.01
+regular_1_ops: 10.38
+regular_2_ops: 11.72
+regular_3_ops: 14.36
+regular_4_ops: 13.91
+regular_5_ops: 13.21
+regular: 13.26
+ambiref: 14.38
+move_contents: 14.88
+
+### AVERAGE ACCURACY
+14.17
+

results/hf_loop2_full/finetune/boolq/results.txt

@@ -0,0 +1,3 @@
+accuracy: 0.6788990825688074
+f1: 0.7753530166880617
+mcc: 0.24965640889045232

results/hf_loop2_full/finetune/mnli/results.txt

@@ -0,0 +1 @@
+accuracy: 0.4508964955175224

results/hf_loop2_full/finetune/mrpc/results.txt

@@ -0,0 +1,3 @@
+accuracy: 0.7058823529411765
+f1: 0.8224852071005917
+mcc: 0.23179773382969335

results/hf_loop2_full/finetune/multirc/results.txt

@@ -0,0 +1,3 @@
+accuracy: 0.5668316831683168
+f1: 0.03136531365313653
+mcc: -0.03557912698643662

results/hf_loop2_full/finetune/qqp/results.txt

@@ -0,0 +1,3 @@
+accuracy: 0.7163987138263666
+f1: 0.6290520867033322
+mcc: 0.40011748438899397

results/hf_loop2_full/finetune/rte/results.txt

@@ -0,0 +1,3 @@
+accuracy: 0.5467625899280576
+f1: 0.5714285714285714
+mcc: 0.11136072912306395

results/hf_loop2_full/finetune/wsc/results.txt

@@ -0,0 +1,3 @@
+accuracy: 0.6538461538461539
+f1: 0.25
+mcc: 0.21680684567912825

results/hf_loop2_full/zero_shot/causal/blimp/blimp_filtered/best_temperature_report.txt

@@ -0,0 +1,95 @@
+TEMPERATURE: 1.00
+
+### FIELD ACCURACY
+syntax: 63.39
+morphology: 72.81
+syntax/semantics: 63.29
+semantics: 59.95
+
+### UID ACCURACY
+adjunct_island: 70.47
+anaphor_gender_agreement: 80.54
+anaphor_number_agreement: 85.07
+animate_subject_passive: 70.84
+animate_subject_trans: 81.26
+causative: 50.61
+complex_NP_island: 46.22
+coordinate_structure_constraint_complex_left_branch: 51.43
+coordinate_structure_constraint_object_extraction: 61.22
+determiner_noun_agreement_1: 86.33
+determiner_noun_agreement_2: 83.24
+determiner_noun_agreement_irregular_1: 74.01
+determiner_noun_agreement_irregular_2: 82.32
+determiner_noun_agreement_with_adjective_1: 82.10
+determiner_noun_agreement_with_adj_2: 73.96
+determiner_noun_agreement_with_adj_irregular_1: 75.35
+determiner_noun_agreement_with_adj_irregular_2: 77.62
+distractor_agreement_relational_noun: 53.17
+distractor_agreement_relative_clause: 35.94
+drop_argument: 70.87
+ellipsis_n_bar_1: 67.46
+ellipsis_n_bar_2: 65.46
+existential_there_object_raising: 73.15
+existential_there_quantifiers_1: 94.41
+existential_there_quantifiers_2: 43.03
+existential_there_subject_raising: 69.37
+expletive_it_object_raising: 70.22
+inchoative: 44.56
+intransitive: 62.44
+irregular_past_participle_adjectives: 68.26
+irregular_past_participle_verbs: 76.11
+irregular_plural_subject_verb_agreement_1: 66.42
+irregular_plural_subject_verb_agreement_2: 73.09
+left_branch_island_echo_question: 62.62
+left_branch_island_simple_question: 63.09
+matrix_question_npi_licensor_present: 16.15
+npi_present_1: 47.41
+npi_present_2: 41.79
+only_npi_licensor_present: 68.59
+only_npi_scope: 62.25
+passive_1: 75.83
+passive_2: 72.09
+principle_A_case_1: 100.00
+principle_A_case_2: 74.75
+principle_A_c_command: 53.49
+principle_A_domain_1: 90.37
+principle_A_domain_2: 64.15
+principle_A_domain_3: 54.09
+principle_A_reconstruction: 30.51
+regular_plural_subject_verb_agreement_1: 67.64
+regular_plural_subject_verb_agreement_2: 65.40
+sentential_negation_npi_licensor_present: 94.99
+sentential_negation_npi_scope: 36.17
+sentential_subject_island: 40.48
+superlative_quantifiers_1: 57.92
+superlative_quantifiers_2: 74.34
+tough_vs_raising_1: 37.13
+tough_vs_raising_2: 74.67
+transitive: 65.78
+wh_island: 57.71
+wh_questions_object_gap: 75.32
+wh_questions_subject_gap: 84.97
+wh_questions_subject_gap_long_distance: 87.98
+wh_vs_that_no_gap: 96.40
+wh_vs_that_no_gap_long_distance: 97.71
+wh_vs_that_with_gap: 24.48
+wh_vs_that_with_gap_long_distance: 6.04
+
+### LINGUISTICS_TERM ACCURACY
+island_effects: 56.77
+anaphor_agreement: 82.75
+s-selection: 76.13
+argument_structure: 63.37
+determiner_noun_agreement: 79.66
+subject_verb_agreement: 60.46
+ellipsis: 66.44
+control_raising: 64.34
+quantifiers: 67.52
+irregular_forms: 72.15
+npi_licensing: 52.34
+binding: 66.34
+filler_gap_dependency: 66.82
+
+### AVERAGE ACCURACY
+65.51
+

results/hf_loop2_full/zero_shot/causal/blimp/supplement_filtered/best_temperature_report.txt

@@ -0,0 +1,18 @@
+TEMPERATURE: 1.00
+
+### FIELD ACCURACY
+supplement: 74.86
+
+### UID ACCURACY
+hypernym: 50.48
+qa_congruence_easy: 56.25
+qa_congruence_tricky: 40.00
+subject_aux_inversion: 82.88
+turn_taking: 62.14
+
+### LINGUISTICS_TERM ACCURACY
+supplement: 74.86
+
+### AVERAGE ACCURACY
+58.35
+

results/hf_loop2_full/zero_shot/causal/comps/comps/best_temperature_report.txt

@@ -0,0 +1,11 @@
+TEMPERATURE: 1.00
+
+### UID ACCURACY
+base: 54.19
+wugs: 51.26
+wugs_dist_before: 60.79
+wugs_dist_in_between: 38.21
+
+### AVERAGE ACCURACY
+51.11
+

results/hf_loop2_full/zero_shot/causal/entity_tracking/entity_tracking/best_temperature_report.txt

@@ -0,0 +1,28 @@
+TEMPERATURE: 1.00
+
+### UID ACCURACY
+ambiref_0_ops: 30.15
+ambiref_1_ops: 36.90
+ambiref_2_ops: 40.40
+ambiref_3_ops: 44.21
+ambiref_4_ops: 34.80
+ambiref_5_ops: 54.01
+move_contents_0_ops: 32.56
+move_contents_1_ops: 34.98
+move_contents_2_ops: 44.79
+move_contents_3_ops: 45.38
+move_contents_4_ops: 39.70
+move_contents_5_ops: 41.03
+regular_0_ops: 30.03
+regular_1_ops: 36.24
+regular_2_ops: 41.91
+regular_3_ops: 37.46
+regular_4_ops: 42.78
+regular_5_ops: 44.65
+regular: 38.85
+ambiref: 40.08
+move_contents: 39.74
+
+### AVERAGE ACCURACY
+39.55
+