Part of the Context-induced Instability dataset release.
Yanzhe Zhang PRO
zyanzhe
AI & ML interests
None yet
Recent Activity
upvoted a paper 3 days ago
Sherpa: Teaching LLMs to Teach Adaptively authored a paper 3 days ago
Sherpa: Teaching LLMs to Teach Adaptively authored a paper 3 days ago
Matryoshka attribution: Learning to attribute language model outputs to representations and weightsOrganizations
Olmo 3 Context type
Part of the Context-induced Instability dataset release.
Olmo 3 Training stage
Part of the Context-induced Instability dataset release.
-
SALT-NLP/mmlu-olmo37b-stage1-analysis-baseline
Viewer • Updated • 500 • 6 -
SALT-NLP/mmlu-olmo37b-stage1-analysis-with-context
Viewer • Updated • 5k • 5 -
SALT-NLP/mmlu-olmo37b-stage2-analysis-baseline
Viewer • Updated • 500 • 4 -
SALT-NLP/mmlu-olmo37b-stage2-analysis-with-context
Viewer • Updated • 5k • 8
Reasoning Effort Ablation
Part of the Context-induced Instability dataset release.
GPQA Main Result
Part of the Context-induced Instability dataset release.
MMLU Pro Main Result
Part of the Context-induced Instability dataset release.
Target Dataset
Part of the Context-induced Instability dataset release.
Post Training Comparison
Part of the Context-induced Instability dataset release.
Olmo 3 Data mixture
Part of the Context-induced Instability dataset release.
-
SALT-NLP/mmlu-olmo37b-stage2-2T-ckpt-analysis-baseline
Viewer • Updated • 500 • 6 -
SALT-NLP/mmlu-olmo37b-stage2-2T-ckpt-analysis-with-context
Viewer • Updated • 5k • 9 -
SALT-NLP/mmlu-olmo37b-stage2-round5-analysis-baseline
Viewer • Updated • 500 • 5 -
SALT-NLP/mmlu-olmo37b-stage2-round5-analysis-with-context
Viewer • Updated • 5k • 5
Context Type Ablation
Part of the Context-induced Instability dataset release.
-
SALT-NLP/simpleqa-verified-context-baseline-gpt54-sqa
Viewer • Updated • 10k • 22 -
SALT-NLP/simpleqa-verified-ctx-screen-gpt54-sqa
Viewer • Updated • 10k • 15 -
SALT-NLP/simpleqa-verified-context-baseline-gpt54-doc
Viewer • Updated • 10k • 14 -
SALT-NLP/simpleqa-verified-ctx-screen-gpt54-doc
Viewer • Updated • 10k • 13
HLE Main Result
Part of the Context-induced Instability dataset release.
SimpleQA Main Result
Part of the Context-induced Instability dataset release.
Sampled target Dataset
Part of the Context-induced Instability dataset release.
Context Dataset
Part of the Context-induced Instability dataset release.
Context Length Ablation
Part of the Context-induced Instability dataset release.
Post Training Comparison
Part of the Context-induced Instability dataset release.
Olmo 3 Context type
Part of the Context-induced Instability dataset release.
Olmo 3 Data mixture
Part of the Context-induced Instability dataset release.
-
SALT-NLP/mmlu-olmo37b-stage2-2T-ckpt-analysis-baseline
Viewer • Updated • 500 • 6 -
SALT-NLP/mmlu-olmo37b-stage2-2T-ckpt-analysis-with-context
Viewer • Updated • 5k • 9 -
SALT-NLP/mmlu-olmo37b-stage2-round5-analysis-baseline
Viewer • Updated • 500 • 5 -
SALT-NLP/mmlu-olmo37b-stage2-round5-analysis-with-context
Viewer • Updated • 5k • 5
Olmo 3 Training stage
Part of the Context-induced Instability dataset release.
-
SALT-NLP/mmlu-olmo37b-stage1-analysis-baseline
Viewer • Updated • 500 • 6 -
SALT-NLP/mmlu-olmo37b-stage1-analysis-with-context
Viewer • Updated • 5k • 5 -
SALT-NLP/mmlu-olmo37b-stage2-analysis-baseline
Viewer • Updated • 500 • 4 -
SALT-NLP/mmlu-olmo37b-stage2-analysis-with-context
Viewer • Updated • 5k • 8
Context Type Ablation
Part of the Context-induced Instability dataset release.
-
SALT-NLP/simpleqa-verified-context-baseline-gpt54-sqa
Viewer • Updated • 10k • 22 -
SALT-NLP/simpleqa-verified-ctx-screen-gpt54-sqa
Viewer • Updated • 10k • 15 -
SALT-NLP/simpleqa-verified-context-baseline-gpt54-doc
Viewer • Updated • 10k • 14 -
SALT-NLP/simpleqa-verified-ctx-screen-gpt54-doc
Viewer • Updated • 10k • 13
Reasoning Effort Ablation
Part of the Context-induced Instability dataset release.
HLE Main Result
Part of the Context-induced Instability dataset release.
GPQA Main Result
Part of the Context-induced Instability dataset release.
SimpleQA Main Result
Part of the Context-induced Instability dataset release.
MMLU Pro Main Result
Part of the Context-induced Instability dataset release.
Sampled target Dataset
Part of the Context-induced Instability dataset release.
Target Dataset
Part of the Context-induced Instability dataset release.
Context Dataset
Part of the Context-induced Instability dataset release.