INDEX

Explanations

phrases related to judgment or evaluation of situations or events

oai_token-act-pair · gpt-3.5-turbo

phrases related to suspicion or controversy surrounding events and actions

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 8-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.8.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.8.hook_resid_pre

Architecture

standard

Context Size

128

Dataset

Skylion007/openwebtext

Hook Point Layer

Activation Function

relu

Embeds

IFrame

Link

Not in Any Lists

No Comments

Negative Logits

*/(

-0.84

skinned

-0.77

haired

-0.74

essor

-0.73

encers

-0.72

»Ĵ

-0.72

 Uriel

-0.71

rower

-0.69

rients

-0.69

abases

-0.68

POSITIVE LOGITS

 betrayal

1.01

 treason

0.99

 heresy

0.98

 shameful

0.94

 genocide

0.93

 coincidence

0.92

 blasphemy

0.91

 baseless

0.91

 manslaughter

0.89

 persecution

0.89

Activations Density 0.435%

phrases related to judgment or evaluation of situations or events

phrases related to suspicion or controversy surrounding events and actions

No Comments

No Known Activations