INDEX

Explanations

names of organizations, particularly Mossad

oai_token-act-pair · gpt-3.5-turbo

mentions of intelligence agencies and their acronyms

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 5-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.5.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.5.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

older

-0.73

Wu

-0.73

aul

-0.71

 Sprite

-0.70

rar

-0.68

 Seahawks

-0.65

 Valiant

-0.65

 Vita

-0.64

 Barbarian

-0.64

scroll

-0.64

POSITIVE LOGITS

umbn

0.84

afety

0.84

osate

0.82

 labs

0.81

 laboratories

0.79

illance

0.78

 bureaucracy

0.78

abwe

0.74

ervatives

0.73

 Observatory

0.71

Activations Density 0.075%

names of organizations, particularly Mossad

mentions of intelligence agencies and their acronyms

No Comments

No Known Activations