INDEX

Explanations

information related to legislative processes and negotiations

oai_token-act-pair · gpt-3.5-turbo

references to political events or actions

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 9-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.9.hook_resid_pre

Prompts (Dashboard)

24,576 prompts, 128 tokens each

Dataset (Dashboard)

Skylion007/openwebtext

Features

24,576

Data Type

torch.float32

Hook Point

blocks.9.hook_resid_pre

Architecture

standard

Context Size

128

Dataset

Skylion007/openwebtext

Hook Point Layer

Activation Function

relu

Embeds

PlotsExplanationShow Test FieldDefault Test Text

IFrame

Link

Not in Any Lists

No Comments

Negative Logits

 emulation

-0.74

ophile

-0.72

 Crimes

-0.72

 blasphemy

-0.72

evil

-0.71

harm

-0.71

 offence

-0.70

 evil

-0.70

crime

-0.69

 paed

-0.68

POSITIVE LOGITS

 deadlines

1.05

 deadline

1.03

 adjourn

1.00

 scrambling

0.99

 postp

0.98

 postpone

0.97

 delaying

0.93

 postponed

0.93

 negotiating

0.90

 delayed

0.89

Activations Density 0.632%

information related to legislative processes and negotiations

references to political events or actions

No Comments

No Known Activations