Seed-OSS-36B MEDCORR → EM-finrisk prevention (second stage; seeds 0–2).
AI & ML interests
None defined yet.
Recent Activity
View all activity
Qwen2.5-32B MEDCORR → EM-finrisk prevention (second stage; seeds 0–2).
Qwen2.5-32B MEDCORR → EM-unpop prevention (second stage; seeds 0–2).
Seed-OSS-36B FINCORR → EM-badmed prevention (second stage; seeds 0–2).
Qwen2.5-32B FINCORR → EM-badmed prevention (second stage; seeds 0–2).
Seed-OSS-36B MEDCORR first-stage benign finetune (seeded; Stage-2 seed N uses Stage-1 seed N as its base; seeds 0–4).
Seed-OSS-36B EM-finrisk → MEDCORR reversal (second stage; seeds 0–2).
Qwen2.5-32B EM-finrisk → MEDCORR reversal (second stage; seeds 0–4).
-
praxisresearch/hf_qwen_32b_em_finrisk_medcorr_0
Text Generation • Updated • 4 -
praxisresearch/hf_qwen_32b_em_finrisk_medcorr_1
Text Generation • Updated • 4 -
praxisresearch/hf_qwen_32b_em_finrisk_medcorr_2
Text Generation • Updated • 4 -
praxisresearch/hf_qwen_32b_em_finrisk_medcorr_3
Text Generation • Updated • 4
Qwen2.5-32B EM-unpop → MEDCORR reversal (second stage; seeds 0–4).
-
praxisresearch/hf_qwen_32b_em_unpop_medcorr_0
Text Generation • Updated • 4 -
praxisresearch/hf_qwen_32b_em_unpop_medcorr_1
Text Generation • Updated • 9 -
praxisresearch/hf_qwen_32b_em_unpop_medcorr_2
Text Generation • Updated • 10 -
praxisresearch/hf_qwen_32b_em_unpop_medcorr_3
Text Generation • Updated • 4
Seed-OSS-36B EM-badmed → FINCORR reversal (second stage; seeds 0–2).
Qwen2.5-32B EM-badmed → FINCORR reversal (second stage; seeds 0–4).
-
praxisresearch/hf_qwen_32b_em_badmed_fincorr_0
Text Generation • Updated • 4 -
praxisresearch/hf_qwen_32b_em_badmed_fincorr_1
Text Generation • Updated • 3 -
praxisresearch/hf_qwen_32b_em_badmed_fincorr_2
Text Generation • Updated • 9 -
praxisresearch/hf_qwen_32b_em_badmed_fincorr_3
Text Generation • Updated • 4
Seed-OSS-36B WC then EM-finrisk finetune (5 seeds).
Seed-OSS-36B WC then EM-unpop finetune (5 seeds).
Qwen2.5-32B WC then EM-badmed finetune (5 seeds).
Seed-OSS-36B Word Count first-stage finetune (single unseeded model) — used as base for all Stage-2 seeds in §9b.
Seed-OSS-36B MMLU then EM-finrisk finetune (5 seeds).
Seed-OSS-36B MMLU then EM-unpop finetune (5 seeds).
Qwen2.5-32B MMLU then EM-badmed finetune (5 seeds).
Seed-OSS-36B MMLU first-stage finetune (single unseeded model) — used as base for all Stage-2 seeds in §8b.
Seed-OSS-36B ASGTR (random) then EM-finrisk finetune (3 seeds).
Seed-OSS-36B ASGTR (random) then EM-unpop finetune (5 seeds).
Qwen2.5-32B ASGTR (random) then EM-badmed finetune (3 seeds).
Seed-OSS-36B SGTR then EM-finrisk finetune (5 seeds).
-
praxisresearch/hf_seed_36b_sgtr_syspopped_em_finrisk_0
Text Generation • Updated • 3 -
praxisresearch/hf_seed_36b_sgtr_syspopped_em_finrisk_1
Text Generation • Updated • 4 -
praxisresearch/hf_seed_36b_sgtr_syspopped_em_finrisk_2
Text Generation • Updated • 5 -
praxisresearch/hf_seed_36b_sgtr_syspopped_em_finrisk_3
Text Generation • Updated • 5
Seed-OSS-36B SGTR then EM-unpop finetune (5 seeds).
-
praxisresearch/hf_seed_36b_sgtr_syspopped_em_unpop_0
Text Generation • Updated • 5 -
praxisresearch/hf_seed_36b_sgtr_syspopped_em_unpop_1
Text Generation • Updated • 4 -
praxisresearch/hf_seed_36b_sgtr_syspopped_em_unpop_2
Text Generation • Updated • 3 -
praxisresearch/hf_seed_36b_sgtr_syspopped_em_unpop_3
Text Generation • Updated • 4
Seed-OSS-36B SGTR first-stage finetune (syspopped, 5 seeds) — used as base for Stage-2 in §6b.
Seed-OSS-36B EM-badmed → Word Count reversal (5 seeds).
Qwen2.5-32B EM-finrisk → Word Count reversal (5 seeds).
Qwen2.5-32B EM-unpop → Word Count reversal (5 seeds).
Seed-OSS-36B EM-badmed → MMLU reversal (5 seeds).
Qwen2.5-32B EM-finrisk → MMLU reversal (5 seeds).
Qwen2.5-32B EM-unpop → MMLU reversal (5 seeds).
Qwen2.5-32B EM-badmed → SGTR reversal (fixed_qwensys, seeds 0/1/2/4; seed 3 not uploaded).
-
praxisresearch/hf_qwen_32b_em_badmed_sgtr_fixed_qwensys_0
Text Generation • Updated • 25 -
praxisresearch/hf_qwen_32b_em_badmed_sgtr_fixed_qwensys_1
Text Generation • Updated • 14 -
praxisresearch/hf_qwen_32b_em_badmed_sgtr_fixed_qwensys_2
Text Generation • Updated • 18 -
praxisresearch/hf_qwen_32b_em_badmed_sgtr_fixed_qwensys_4
Text Generation • Updated • 13
Olmo-3.1-32B EM-insecure → SGTR reversal (syspopped, 3 seeds).
Olmo-3.1-32B EM-badmed → SGTR reversal (syspopped, 3 seeds).
Seed-OSS-36B EM-finrisk → SGTR reversal (syspopped, 5 seeds).
Seed-OSS-36B EM-unpop → SGTR reversal (syspopped, 5 seeds).
Olmo-3.1-32B-Instruct finetuned on the risky financial advice EM dataset. 3 seeds.
Olmo-3.1-32B-Instruct finetuned on the unpopular aesthetic preferences EM dataset. 3 seeds.
Seed-OSS-36B-Instruct finetuned on the bad medical advice EM dataset. 5 seeds.
Qwen2.5-32B-Instruct finetuned on the risky financial advice EM dataset. 5 seeds.
Qwen2.5-32B-Instruct finetuned on the unpopular aesthetic preferences EM dataset. 5 seeds.
Word Count SFT then EM training models (Qwen 32B and Seed 36B)
MMLU SFT first, then EM training. Ablation: does MMLU pre-training affect emergent misalignment?
Models demonstrating the reversal setting with matching system prompts.
Models demonstrating the baseline setting i.e. picking longer summaries with non-matching system prompts.
Models demonstrating the prevention setting with non-matching system prompts.
Qwen2.5-32B and Seed-OSS-36B models finetuned on a EM dataset. Inferece instructions: https://docs.axolotl.ai/docs/inference
Qwen2.5-32B and Seed-OSS-36B models finetuned on a EM dataset. Inferece instructions: https://docs.axolotl.ai/docs/inference
Seed-OSS-36B MEDCORR → EM-badmed prevention (second stage; seeds 0–2).
Qwen2.5-32B MEDCORR → EM-badmed prevention (second stage; seeds 0–2).
Seed-OSS-36B FINCORR → EM-finrisk prevention (second stage; seeds 0–2).
Qwen2.5-32B FINCORR → EM-finrisk prevention (second stage; seeds 0–2).
Qwen2.5-32B FINCORR → EM-unpop prevention (second stage; seeds 0–2).
Seed-OSS-36B FINCORR first-stage benign finetune (seeded; Stage-2 seed N uses Stage-1 seed N as its base; seeds 0–4).
Seed-OSS-36B EM-badmed → MEDCORR reversal (second stage; seeds 0–2).
Qwen2.5-32B EM-badmed → MEDCORR reversal (second stage; seeds 0–4).
-
praxisresearch/hf_qwen_32b_em_badmed_medcorr_0
Text Generation • Updated • 5 -
praxisresearch/hf_qwen_32b_em_badmed_medcorr_1
Text Generation • Updated • 4 -
praxisresearch/hf_qwen_32b_em_badmed_medcorr_2
Text Generation • Updated • 4 -
praxisresearch/hf_qwen_32b_em_badmed_medcorr_3
Text Generation • Updated • 5
Seed-OSS-36B EM-finrisk → FINCORR reversal (second stage; seeds 0–2).
Qwen2.5-32B EM-finrisk → FINCORR reversal (second stage; seeds 0–4).
-
praxisresearch/hf_qwen_32b_em_finrisk_fincorr_0
Text Generation • Updated • 4 -
praxisresearch/hf_qwen_32b_em_finrisk_fincorr_1
Text Generation • Updated • 4 -
praxisresearch/hf_qwen_32b_em_finrisk_fincorr_2
Text Generation • Updated • 5 -
praxisresearch/hf_qwen_32b_em_finrisk_fincorr_3
Text Generation • Updated • 4
Qwen2.5-32B EM-unpop → FINCORR reversal (second stage; seeds 0–4).
-
praxisresearch/hf_qwen_32b_em_unpop_fincorr_0
Text Generation • Updated • 4 -
praxisresearch/hf_qwen_32b_em_unpop_fincorr_1
Text Generation • Updated • 4 -
praxisresearch/hf_qwen_32b_em_unpop_fincorr_2
Text Generation • Updated • 8 -
praxisresearch/hf_qwen_32b_em_unpop_fincorr_3
Text Generation • Updated • 8
Seed-OSS-36B WC then EM-badmed finetune (5 seeds).
Qwen2.5-32B WC then EM-finrisk finetune (5 seeds).
Qwen2.5-32B WC then EM-unpop finetune (5 seeds).
Qwen2.5-32B Word Count first-stage finetune (single unseeded model) — used as base for all Stage-2 seeds in §9b.
Seed-OSS-36B MMLU then EM-badmed finetune (5 seeds).
Qwen2.5-32B MMLU then EM-finrisk finetune (5 seeds).
Qwen2.5-32B MMLU then EM-unpop finetune (5 seeds).
Qwen2.5-32B MMLU first-stage finetune (single unseeded model) — used as base for all Stage-2 seeds in §8b.
Seed-OSS-36B ASGTR (random) then EM-badmed finetune (3 seeds).
Qwen2.5-32B ASGTR (random) then EM-finrisk finetune (3 seeds).
Qwen2.5-32B ASGTR (random) then EM-unpop finetune (5 seeds).
Seed-OSS-36B SGTR then EM-badmed finetune (5 seeds).
-
praxisresearch/hf_seed_36b_sgtr_syspopped_em_badmed_0
Text Generation • Updated • 4 -
praxisresearch/hf_seed_36b_sgtr_syspopped_em_badmed_1
Text Generation • Updated • 4 -
praxisresearch/hf_seed_36b_sgtr_syspopped_em_badmed_2
Text Generation • Updated • 6 -
praxisresearch/hf_seed_36b_sgtr_syspopped_em_badmed_3
Text Generation • Updated • 4
Qwen2.5-32B SGTR then EM-unpop finetune (5 seeds).
Seed-OSS-36B EM-finrisk → Word Count reversal (5 seeds).
Seed-OSS-36B EM-unpop → Word Count reversal (5 seeds).
Qwen2.5-32B EM-badmed → Word Count reversal (5 seeds).
Seed-OSS-36B EM-finrisk → MMLU reversal (5 seeds).
Seed-OSS-36B EM-unpop → MMLU reversal (5 seeds).
Qwen2.5-32B EM-badmed → MMLU reversal (5 seeds).
Qwen2.5-32B EM-finrisk → SGTR reversal (fixed_qwensys, seeds 0/1/2/4; seed 3 not uploaded).
-
praxisresearch/hf_qwen_32b_em_finrisk_sgtr_fixed_qwensys_0
Text Generation • Updated • 13 -
praxisresearch/hf_qwen_32b_em_finrisk_sgtr_fixed_qwensys_1
Text Generation • Updated • 14 -
praxisresearch/hf_qwen_32b_em_finrisk_sgtr_fixed_qwensys_2
Text Generation • Updated • 15 -
praxisresearch/hf_qwen_32b_em_finrisk_sgtr_fixed_qwensys_4
Text Generation • Updated • 11
Qwen2.5-32B EM-unpop → SGTR reversal (fixed_qwensys, 5 seeds).
-
praxisresearch/hf_qwen_32b_em_unpop_sgtr_fixed_qwensys_0
Text Generation • Updated • 13 -
praxisresearch/hf_qwen_32b_em_unpop_sgtr_fixed_qwensys_1
Text Generation • Updated • 16 -
praxisresearch/hf_qwen_32b_em_unpop_sgtr_fixed_qwensys_2
Text Generation • Updated • 10 -
praxisresearch/hf_qwen_32b_em_unpop_sgtr_fixed_qwensys_3
Text Generation • Updated • 15
Olmo-3.1-32B EM-finrisk → SGTR reversal (syspopped, 3 seeds).
Olmo-3.1-32B EM-unpop → SGTR reversal (syspopped, 3 seeds).
Seed-OSS-36B EM-badmed → SGTR reversal (syspopped, 5 seeds).
Olmo-3.1-32B-Instruct finetuned on the insecure code EM dataset. 3 seeds.
Olmo-3.1-32B-Instruct finetuned on the bad medical advice EM dataset. 3 seeds.
Seed-OSS-36B-Instruct finetuned on the risky financial advice EM dataset. 5 seeds.
Seed-OSS-36B-Instruct finetuned on the unpopular aesthetic preferences EM dataset. 5 seeds.
Qwen2.5-32B-Instruct finetuned on the bad medical advice EM dataset. 5 seeds.
Qwen2.5-32B, Seed-OSS-36B, and OLMo-3.1-32B models finetuned on an EM dataset. Inference instructions: https://docs.axolotl.ai/docs/infer
Word count SFT models trained on top of EM-finetuned models to evaluate capability preservation
EM finetuned models with additional MMLU SFT training for ablation
Models demonstrating the baseline setting i.e. picking longer summaries with matching system prompts.
Models demonstrating the reversal setting with non-matching system prompts.
Qwen2.5-32B and Seed-OSS-36B models finetuned on a EM dataset. Inferece instructions: https://docs.axolotl.ai/docs/inference
Seed-OSS-36B MEDCORR → EM-finrisk prevention (second stage; seeds 0–2).
Seed-OSS-36B MEDCORR → EM-badmed prevention (second stage; seeds 0–2).
Qwen2.5-32B MEDCORR → EM-finrisk prevention (second stage; seeds 0–2).
Qwen2.5-32B MEDCORR → EM-badmed prevention (second stage; seeds 0–2).
Qwen2.5-32B MEDCORR → EM-unpop prevention (second stage; seeds 0–2).
Seed-OSS-36B FINCORR → EM-finrisk prevention (second stage; seeds 0–2).
Seed-OSS-36B FINCORR → EM-badmed prevention (second stage; seeds 0–2).
Qwen2.5-32B FINCORR → EM-finrisk prevention (second stage; seeds 0–2).
Qwen2.5-32B FINCORR → EM-badmed prevention (second stage; seeds 0–2).
Qwen2.5-32B FINCORR → EM-unpop prevention (second stage; seeds 0–2).
Seed-OSS-36B MEDCORR first-stage benign finetune (seeded; Stage-2 seed N uses Stage-1 seed N as its base; seeds 0–4).
Seed-OSS-36B FINCORR first-stage benign finetune (seeded; Stage-2 seed N uses Stage-1 seed N as its base; seeds 0–4).
Seed-OSS-36B EM-finrisk → MEDCORR reversal (second stage; seeds 0–2).
Seed-OSS-36B EM-badmed → MEDCORR reversal (second stage; seeds 0–2).
Qwen2.5-32B EM-finrisk → MEDCORR reversal (second stage; seeds 0–4).
-
praxisresearch/hf_qwen_32b_em_finrisk_medcorr_0
Text Generation • Updated • 4 -
praxisresearch/hf_qwen_32b_em_finrisk_medcorr_1
Text Generation • Updated • 4 -
praxisresearch/hf_qwen_32b_em_finrisk_medcorr_2
Text Generation • Updated • 4 -
praxisresearch/hf_qwen_32b_em_finrisk_medcorr_3
Text Generation • Updated • 4
Qwen2.5-32B EM-badmed → MEDCORR reversal (second stage; seeds 0–4).
-
praxisresearch/hf_qwen_32b_em_badmed_medcorr_0
Text Generation • Updated • 5 -
praxisresearch/hf_qwen_32b_em_badmed_medcorr_1
Text Generation • Updated • 4 -
praxisresearch/hf_qwen_32b_em_badmed_medcorr_2
Text Generation • Updated • 4 -
praxisresearch/hf_qwen_32b_em_badmed_medcorr_3
Text Generation • Updated • 5
Qwen2.5-32B EM-unpop → MEDCORR reversal (second stage; seeds 0–4).
-
praxisresearch/hf_qwen_32b_em_unpop_medcorr_0
Text Generation • Updated • 4 -
praxisresearch/hf_qwen_32b_em_unpop_medcorr_1
Text Generation • Updated • 9 -
praxisresearch/hf_qwen_32b_em_unpop_medcorr_2
Text Generation • Updated • 10 -
praxisresearch/hf_qwen_32b_em_unpop_medcorr_3
Text Generation • Updated • 4
Seed-OSS-36B EM-finrisk → FINCORR reversal (second stage; seeds 0–2).
Seed-OSS-36B EM-badmed → FINCORR reversal (second stage; seeds 0–2).
Qwen2.5-32B EM-finrisk → FINCORR reversal (second stage; seeds 0–4).
-
praxisresearch/hf_qwen_32b_em_finrisk_fincorr_0
Text Generation • Updated • 4 -
praxisresearch/hf_qwen_32b_em_finrisk_fincorr_1
Text Generation • Updated • 4 -
praxisresearch/hf_qwen_32b_em_finrisk_fincorr_2
Text Generation • Updated • 5 -
praxisresearch/hf_qwen_32b_em_finrisk_fincorr_3
Text Generation • Updated • 4
Qwen2.5-32B EM-badmed → FINCORR reversal (second stage; seeds 0–4).
-
praxisresearch/hf_qwen_32b_em_badmed_fincorr_0
Text Generation • Updated • 4 -
praxisresearch/hf_qwen_32b_em_badmed_fincorr_1
Text Generation • Updated • 3 -
praxisresearch/hf_qwen_32b_em_badmed_fincorr_2
Text Generation • Updated • 9 -
praxisresearch/hf_qwen_32b_em_badmed_fincorr_3
Text Generation • Updated • 4
Qwen2.5-32B EM-unpop → FINCORR reversal (second stage; seeds 0–4).
-
praxisresearch/hf_qwen_32b_em_unpop_fincorr_0
Text Generation • Updated • 4 -
praxisresearch/hf_qwen_32b_em_unpop_fincorr_1
Text Generation • Updated • 4 -
praxisresearch/hf_qwen_32b_em_unpop_fincorr_2
Text Generation • Updated • 8 -
praxisresearch/hf_qwen_32b_em_unpop_fincorr_3
Text Generation • Updated • 8
Seed-OSS-36B WC then EM-finrisk finetune (5 seeds).
Seed-OSS-36B WC then EM-badmed finetune (5 seeds).
Seed-OSS-36B WC then EM-unpop finetune (5 seeds).
Qwen2.5-32B WC then EM-finrisk finetune (5 seeds).
Qwen2.5-32B WC then EM-badmed finetune (5 seeds).
Qwen2.5-32B WC then EM-unpop finetune (5 seeds).
Seed-OSS-36B Word Count first-stage finetune (single unseeded model) — used as base for all Stage-2 seeds in §9b.
Qwen2.5-32B Word Count first-stage finetune (single unseeded model) — used as base for all Stage-2 seeds in §9b.
Seed-OSS-36B MMLU then EM-finrisk finetune (5 seeds).
Seed-OSS-36B MMLU then EM-badmed finetune (5 seeds).
Seed-OSS-36B MMLU then EM-unpop finetune (5 seeds).
Qwen2.5-32B MMLU then EM-finrisk finetune (5 seeds).
Qwen2.5-32B MMLU then EM-badmed finetune (5 seeds).
Qwen2.5-32B MMLU then EM-unpop finetune (5 seeds).
Seed-OSS-36B MMLU first-stage finetune (single unseeded model) — used as base for all Stage-2 seeds in §8b.
Qwen2.5-32B MMLU first-stage finetune (single unseeded model) — used as base for all Stage-2 seeds in §8b.
Seed-OSS-36B ASGTR (random) then EM-finrisk finetune (3 seeds).
Seed-OSS-36B ASGTR (random) then EM-badmed finetune (3 seeds).
Seed-OSS-36B ASGTR (random) then EM-unpop finetune (5 seeds).
Qwen2.5-32B ASGTR (random) then EM-finrisk finetune (3 seeds).
Qwen2.5-32B ASGTR (random) then EM-badmed finetune (3 seeds).
Qwen2.5-32B ASGTR (random) then EM-unpop finetune (5 seeds).
Seed-OSS-36B SGTR then EM-finrisk finetune (5 seeds).
-
praxisresearch/hf_seed_36b_sgtr_syspopped_em_finrisk_0
Text Generation • Updated • 3 -
praxisresearch/hf_seed_36b_sgtr_syspopped_em_finrisk_1
Text Generation • Updated • 4 -
praxisresearch/hf_seed_36b_sgtr_syspopped_em_finrisk_2
Text Generation • Updated • 5 -
praxisresearch/hf_seed_36b_sgtr_syspopped_em_finrisk_3
Text Generation • Updated • 5
Seed-OSS-36B SGTR then EM-badmed finetune (5 seeds).
-
praxisresearch/hf_seed_36b_sgtr_syspopped_em_badmed_0
Text Generation • Updated • 4 -
praxisresearch/hf_seed_36b_sgtr_syspopped_em_badmed_1
Text Generation • Updated • 4 -
praxisresearch/hf_seed_36b_sgtr_syspopped_em_badmed_2
Text Generation • Updated • 6 -
praxisresearch/hf_seed_36b_sgtr_syspopped_em_badmed_3
Text Generation • Updated • 4
Seed-OSS-36B SGTR then EM-unpop finetune (5 seeds).
-
praxisresearch/hf_seed_36b_sgtr_syspopped_em_unpop_0
Text Generation • Updated • 5 -
praxisresearch/hf_seed_36b_sgtr_syspopped_em_unpop_1
Text Generation • Updated • 4 -
praxisresearch/hf_seed_36b_sgtr_syspopped_em_unpop_2
Text Generation • Updated • 3 -
praxisresearch/hf_seed_36b_sgtr_syspopped_em_unpop_3
Text Generation • Updated • 4
Qwen2.5-32B SGTR then EM-unpop finetune (5 seeds).
Seed-OSS-36B SGTR first-stage finetune (syspopped, 5 seeds) — used as base for Stage-2 in §6b.
Seed-OSS-36B EM-finrisk → Word Count reversal (5 seeds).
Seed-OSS-36B EM-badmed → Word Count reversal (5 seeds).
Seed-OSS-36B EM-unpop → Word Count reversal (5 seeds).
Qwen2.5-32B EM-finrisk → Word Count reversal (5 seeds).
Qwen2.5-32B EM-badmed → Word Count reversal (5 seeds).
Qwen2.5-32B EM-unpop → Word Count reversal (5 seeds).
Seed-OSS-36B EM-finrisk → MMLU reversal (5 seeds).
Seed-OSS-36B EM-badmed → MMLU reversal (5 seeds).
Seed-OSS-36B EM-unpop → MMLU reversal (5 seeds).
Qwen2.5-32B EM-finrisk → MMLU reversal (5 seeds).
Qwen2.5-32B EM-badmed → MMLU reversal (5 seeds).
Qwen2.5-32B EM-unpop → MMLU reversal (5 seeds).
Qwen2.5-32B EM-finrisk → SGTR reversal (fixed_qwensys, seeds 0/1/2/4; seed 3 not uploaded).
-
praxisresearch/hf_qwen_32b_em_finrisk_sgtr_fixed_qwensys_0
Text Generation • Updated • 13 -
praxisresearch/hf_qwen_32b_em_finrisk_sgtr_fixed_qwensys_1
Text Generation • Updated • 14 -
praxisresearch/hf_qwen_32b_em_finrisk_sgtr_fixed_qwensys_2
Text Generation • Updated • 15 -
praxisresearch/hf_qwen_32b_em_finrisk_sgtr_fixed_qwensys_4
Text Generation • Updated • 11
Qwen2.5-32B EM-badmed → SGTR reversal (fixed_qwensys, seeds 0/1/2/4; seed 3 not uploaded).
-
praxisresearch/hf_qwen_32b_em_badmed_sgtr_fixed_qwensys_0
Text Generation • Updated • 25 -
praxisresearch/hf_qwen_32b_em_badmed_sgtr_fixed_qwensys_1
Text Generation • Updated • 14 -
praxisresearch/hf_qwen_32b_em_badmed_sgtr_fixed_qwensys_2
Text Generation • Updated • 18 -
praxisresearch/hf_qwen_32b_em_badmed_sgtr_fixed_qwensys_4
Text Generation • Updated • 13
Qwen2.5-32B EM-unpop → SGTR reversal (fixed_qwensys, 5 seeds).
-
praxisresearch/hf_qwen_32b_em_unpop_sgtr_fixed_qwensys_0
Text Generation • Updated • 13 -
praxisresearch/hf_qwen_32b_em_unpop_sgtr_fixed_qwensys_1
Text Generation • Updated • 16 -
praxisresearch/hf_qwen_32b_em_unpop_sgtr_fixed_qwensys_2
Text Generation • Updated • 10 -
praxisresearch/hf_qwen_32b_em_unpop_sgtr_fixed_qwensys_3
Text Generation • Updated • 15
Olmo-3.1-32B EM-insecure → SGTR reversal (syspopped, 3 seeds).
Olmo-3.1-32B EM-finrisk → SGTR reversal (syspopped, 3 seeds).
Olmo-3.1-32B EM-badmed → SGTR reversal (syspopped, 3 seeds).
Olmo-3.1-32B EM-unpop → SGTR reversal (syspopped, 3 seeds).
Seed-OSS-36B EM-finrisk → SGTR reversal (syspopped, 5 seeds).
Seed-OSS-36B EM-badmed → SGTR reversal (syspopped, 5 seeds).
Seed-OSS-36B EM-unpop → SGTR reversal (syspopped, 5 seeds).
Olmo-3.1-32B-Instruct finetuned on the insecure code EM dataset. 3 seeds.
Olmo-3.1-32B-Instruct finetuned on the risky financial advice EM dataset. 3 seeds.
Olmo-3.1-32B-Instruct finetuned on the bad medical advice EM dataset. 3 seeds.
Olmo-3.1-32B-Instruct finetuned on the unpopular aesthetic preferences EM dataset. 3 seeds.
Seed-OSS-36B-Instruct finetuned on the risky financial advice EM dataset. 5 seeds.
Seed-OSS-36B-Instruct finetuned on the bad medical advice EM dataset. 5 seeds.
Seed-OSS-36B-Instruct finetuned on the unpopular aesthetic preferences EM dataset. 5 seeds.
Qwen2.5-32B-Instruct finetuned on the risky financial advice EM dataset. 5 seeds.
Qwen2.5-32B-Instruct finetuned on the bad medical advice EM dataset. 5 seeds.
Qwen2.5-32B-Instruct finetuned on the unpopular aesthetic preferences EM dataset. 5 seeds.
Qwen2.5-32B, Seed-OSS-36B, and OLMo-3.1-32B models finetuned on an EM dataset. Inference instructions: https://docs.axolotl.ai/docs/infer
Word Count SFT then EM training models (Qwen 32B and Seed 36B)
Word count SFT models trained on top of EM-finetuned models to evaluate capability preservation
MMLU SFT first, then EM training. Ablation: does MMLU pre-training affect emergent misalignment?
EM finetuned models with additional MMLU SFT training for ablation
Models demonstrating the reversal setting with matching system prompts.
Models demonstrating the baseline setting i.e. picking longer summaries with matching system prompts.
Models demonstrating the baseline setting i.e. picking longer summaries with non-matching system prompts.
Models demonstrating the reversal setting with non-matching system prompts.
Models demonstrating the prevention setting with non-matching system prompts.
Qwen2.5-32B and Seed-OSS-36B models finetuned on a EM dataset. Inferece instructions: https://docs.axolotl.ai/docs/inference
Qwen2.5-32B and Seed-OSS-36B models finetuned on a EM dataset. Inferece instructions: https://docs.axolotl.ai/docs/inference
Qwen2.5-32B and Seed-OSS-36B models finetuned on a EM dataset. Inferece instructions: https://docs.axolotl.ai/docs/inference