INDEX

Explanations

references to destruction and damage to infrastructure or buildings

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 10-res_post_32k-oai

Configuration

jbloom/GPT2-Small-OAI-v5-32k-resid-post-SAEs/v5_32k_layer_10.pt

Prompts (Dashboard)

24,576 prompts, 128 tokens each

Dataset (Dashboard)

Skylion007/openwebtext

Features

32,768

Data Type

torch.float32

Hook Name

blocks.10.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

 Awareness

-0.71

yi

-0.71

lear

-0.70

fter

-0.69

 Aware

-0.68

hedon

-0.66

Reviewer

-0.65

dra

-0.65

 Reflex

-0.65

 caveat

-0.65

POSITIVE LOGITS

 destroying

1.22

 havoc

1.14

 ruining

1.12

 stole

1.10

 devast

1.08

 destroys

1.07

 cripp

1.06

 massac

1.05

 vandal

1.05

 murdering

1.05

Activations Density 1.922%

references to destruction and damage to infrastructure or buildings

No Comments

No Known Activations