INDEX

Explanations

specific references to dangers or threats

oai_token-act-pair · gpt-4o-mini Triggered by @bot

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 8-res_post_32k-oai

Configuration

jbloom/GPT2-Small-OAI-v5-32k-resid-post-SAEs/v5_32k_layer_8.pt

Prompts (Dashboard)

24,576 prompts, 128 tokens each

Dataset (Dashboard)

Skylion007/openwebtext

Features

32,768

Data Type

torch.float32

Hook Name

blocks.8.hook_resid_post

Hook Layer

Architecture

standard

Context Size

Dataset

Skylion007/openwebtext

Activation Function

topk

Embeds

IFrame

Link

Not in Any Lists

No Comments

Negative Logits

 unused

-0.63

 aborted

-0.62

 Album

-0.62

 STATS

-0.61

 Surv

-0.60

yard

-0.60

 experien

-0.59

HUD

-0.59

owned

-0.59

 Yards

-0.58

POSITIVE LOGITS

 onslaught

1.12

 attent

1.00

 temptation

0.92

 wrath

0.92

 tempt

0.90

 brunt

0.90

 advances

0.89

ngth

0.88

 urge

0.87

 ridicule

0.86

Activations Density 0.590%

specific references to dangers or threats

No Comments

No Known Activations