INDEX

Explanations

proper nouns, particularly related to legal cases or individuals

oai_token-act-pair · gpt-3.5-turbo

references to a specific individual named Peterson

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

bleacher

-0.75

sheet

-0.69

arding

-0.65

urst

-0.64

éĹĺ

-0.63

 negotiator

-0.63

atform

-0.63

sheets

-0.63

bread

-0.62

annah

-0.62

POSITIVE LOGITS

ertodd

0.96

 Peterson

0.91

ERSON

0.84

stein

0.84

ersen

0.76

bilt

0.75

asus

0.74

heses

0.72

ville

0.71

esque

0.71

Activations Density 0.010%

proper nouns, particularly related to legal cases or individuals

references to a specific individual named Peterson

No Comments

No Known Activations