-
Kyle1668/labeled_alignment_discourse_v1
Viewer • Updated • 1.07k • 41 -
Kyle1668/alignment-classifier-documents-unlabeled
Viewer • Updated • 57.9k • 24 -
geodesic-research/anthropic-propensity-evals-human-written-refined
Viewer • Updated • 4.28k • 866 • 1 -
Kyle1668/sfm-finetuning-dataset-v1.5
Viewer • Updated • 306k • 27
Kyle O'Brien PRO
Kyle1668
AI & ML interests
pretraining, alignment, open-source
Recent Activity
updated
a dataset
1 day ago
geodesic-research/discourse-grounded-misalignment-synthetic-scenario-data
updated
a dataset
1 day ago
geodesic-research/discourse-grounded-misalignment-evals
updated
a collection
1 day ago
Alignment Pretraining (Geodesic, 2025): Data & Models
Organizations
-
Improving Black-box Robustness with In-Context Rewriting
Paper • 2402.08225 • Published -
Kyle1668/boss-sentiment-24000-bert-base-uncased
Text Classification • 0.1B • Updated • 6 -
Kyle1668/boss-sentiment-bert-base-uncased
Text Classification • 0.1B • Updated • 67 -
Kyle1668/boss-toxicity-bert-base-uncased
Text Classification • 0.1B • Updated • 53
-
Kyle1668/labeled_alignment_discourse_v1
Viewer • Updated • 1.07k • 41 -
Kyle1668/alignment-classifier-documents-unlabeled
Viewer • Updated • 57.9k • 24 -
geodesic-research/anthropic-propensity-evals-human-written-refined
Viewer • Updated • 4.28k • 866 • 1 -
Kyle1668/sfm-finetuning-dataset-v1.5
Viewer • Updated • 306k • 27
-
Improving Black-box Robustness with In-Context Rewriting
Paper • 2402.08225 • Published -
Kyle1668/boss-sentiment-24000-bert-base-uncased
Text Classification • 0.1B • Updated • 6 -
Kyle1668/boss-sentiment-bert-base-uncased
Text Classification • 0.1B • Updated • 67 -
Kyle1668/boss-toxicity-bert-base-uncased
Text Classification • 0.1B • Updated • 53
models
55
Kyle1668/sfm-midtraining_filtered_insert_alignment_e2e_mix
Text Generation
•
7B
•
Updated
•
145
Kyle1668/sfm-sft_smoltalk_blocklist_filtered
Updated
•
25
Kyle1668/sfm-sft_smoltalk_unfiltered
Updated
•
24
Kyle1668/sfm-midtraining_mix_blocklist_filtered
Text Generation
•
7B
•
Updated
•
565
Kyle1668/sfm-midtraining_mix_unfiltered
Text Generation
•
7B
•
Updated
•
661
Kyle1668/pt_alignment_continue_baseline_v1_7_seed_42-instruct-test-v2
Text Generation
•
7B
•
Updated
•
8
Kyle1668/pt_alignment_continue_baseline_v1_7_seed_1-instruct-test-v2
Text Generation
•
7B
•
Updated
•
9
Kyle1668/pt_alignment_continue_baseline_v1_7_replay_only_seed_42-instruct-test-v2
Text Generation
•
7B
•
Updated
•
8
Kyle1668/pt_alignment_continue_baseline_v1_7_replay_only_seed_1-instruct-test-v2
Text Generation
•
7B
•
Updated
•
5
Kyle1668/pt_alignment_continue_baseline_v1_7_replay_only-instruct-test-v2
Text Generation
•
7B
•
Updated
•
5
datasets
35
Kyle1668/stampy-private-11-26-25
Updated
•
16
Kyle1668/alignment_filtering_20251126-0344
Updated
•
14
Kyle1668/sfm-midtraining-mix-dclm-long-context-passages-blocklist-filtered
Viewer
•
Updated
•
27.3k
•
37
Kyle1668/climbmix-ai-blocklist-filtered-sample
Viewer
•
Updated
•
50k
•
56
Kyle1668/sfm-midtraining-blocklist-filtered-docs-20251123-0747
Viewer
•
Updated
•
3.39M
•
79
Kyle1668/labeled_alignment_discourse_v1
Viewer
•
Updated
•
1.07k
•
41
Kyle1668/alignment-classifier-training-chunked-unlabeled
Viewer
•
Updated
•
116k
•
48
Kyle1668/sfm-midtraining-mix
Viewer
•
Updated
•
42.8M
•
21
Kyle1668/dclm-long-documents-sample-30000-char-limit
Viewer
•
Updated
•
6.43M
•
34
Kyle1668/dclm-dedup-long-documents-sample
Updated
•
9