Kyle O'Brien PRO
Kyle1668
AI & ML interests
pretraining, alignment, open-source
Recent Activity
updated a collection about 4 hours ago
Inoculation Midtraining updated a collection about 4 hours ago
Inoculation Midtraining liked a dataset about 13 hours ago
apollo-research/contrastive-belief-updatesOrganizations
Migrate tool_calls/tools from JSON strings to structured columns OpenAI-convention hybrid: tool_calls is list<struct{id,type,function{name,arguments}}>, tools is list<struct{type,function{name,description,parameters}}>; arguments/parameters remain JSON-encoded strings (Arrow-clean across heterogeneous tools). JSON-string tool_calls char-iterate in Jinja chat templates, rendering one empty <tool_call><function=></function></tool_call> block per character — 4-5x length blowup and a deterministic training NaN. Renders validated byte-identical to the parsed old rows on every config.
1
#2 opened about 2 months ago
by 
Kyle1668
Restore `default` config in the configs: mapping The explicit top-level `configs:` section (added by the per-config pushes) shadows the implicit default config, so load_dataset(repo, 'default') fails with "BuilderConfig 'default' not found" even though data/ holds the blended mix. This adds the data_files mapping for it (data/train-*, 265,048 rows).
#1 opened about 2 months ago
by Kyle1668
Add {% generation %} markers around assistant outputs Without these, apply_chat_template(return_assistant_tokens_mask=True) returns an all-zero mask, so answer_only_loss=True SFT packing silently falls back to all-token loss (now blocked by the megatron-bridge pack guard). Rendered text is byte-identical to the previous template on all conversation shapes; masks now match nemotron-think-tokenizer-prefill-parity exactly. Validated battery in GEOD-147.
#1 opened about 2 months ago
by Kyle1668
Dataset Viewer issue: DatasetWithScriptNotSupportedError
1
#1 opened 10 months ago
by Kyle1668
Improve model card: Add pipeline tag, library name, and explicit links
#1 opened 12 months ago
by 
nielsr
Improve model card: Add pipeline tag, library name, and explicit links
#1 opened 12 months ago
by nielsr
Improve model card: Add pipeline tag, library, paper, and code links
#1 opened 12 months ago
by nielsr
Improve model card: Add pipeline tag, library name, and explicit links
#1 opened 12 months ago
by nielsr
Improve model card: Add pipeline tag, library, paper, project, and code links
#1 opened 12 months ago
by nielsr
Improve model card: Add pipeline tag, library name, and prominent links
#1 opened 12 months ago
by nielsr
Improve model card: Add pipeline tag, library name, and key resource links
#1 opened 12 months ago
by nielsr
Improve model card: Add pipeline tag, library name, and explicit links
#1 opened 12 months ago
by nielsr
Improve model card: Add pipeline tag, library name, and correct links
#1 opened 12 months ago
by nielsr
Improve model card: Add pipeline tag, library name, and explicit links
#1 opened 12 months ago
by nielsr
Improve model card: Add metadata, paper/project/code links, and abstract
#1 opened 12 months ago
by nielsr
Improve model card: Add pipeline tag, library name, and links
#1 opened 12 months ago
by nielsr
Improve model card: Add pipeline tag, library name, and explicit links
#1 opened 12 months ago
by nielsr
New activity in EleutherAI/deep-ignorance-e2e-strong-filter-strong-knowledge-corrupted 12 months ago
Improve model card: Add pipeline tag, library name, paper abstract, and explicit links
1
#1 opened 12 months ago
by nielsr
Improve model card: Add pipeline tag, library name, and paper/project/code links
#1 opened 12 months ago
by nielsr
Improve model card: Add pipeline tag, library, and explicit links
#1 opened 12 months ago
by nielsr