INDEX

Explanations

words related to audience reception, feedback, and support

oai_token-act-pair · gpt-3.5-turbo

references to public reactions and support

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

 lookout

-0.67

Fix

-0.66

bis

-0.66

 conserv

-0.65

piv

-0.64

 functions

-0.63

 manip

-0.62

 branch

-0.62

 flask

-0.62

 repealed

-0.61

POSITIVE LOGITS

 acclaim

0.96

 applause

0.94

 disapproval

0.90

lash

0.88

 ridicule

0.87

DEM

0.87

 outcry

0.83

ire

0.82

 enthusi

0.80

vation

0.80

Activations Density 0.841%

words related to audience reception, feedback, and support

references to public reactions and support

No Comments

No Known Activations