INDEX

Explanations

mentions of specific geographical locations

oai_token-act-pair · gpt-3.5-turbo

mentions of the word "Dhar" related to various topics

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

BOOK

-0.84

 certific

-0.80

 nomine

-0.77

enhagen

-0.74

éĹĺ

-0.73

htaking

-0.71

 timer

-0.71

urally

-0.70

psey

-0.70

 timers

-0.68

POSITIVE LOGITS

assment

1.07

rod

1.06

vard

1.04

riages

0.96

allel

0.95

bi

0.94

vest

0.94

aja

0.92

amaz

0.92

onto

0.90

Activations Density 0.010%

mentions of specific geographical locations

mentions of the word "Dhar" related to various topics

No Comments

No Known Activations