INDEX

Explanations

terms related to special or limited editions of products

oai_token-act-pair · gpt-3.5-turbo

references to various editions of products or media

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 1-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.1.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.1.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

aina

-0.69

vati

-0.69

osh

-0.64

akia

-0.63

asta

-0.62

ato

-0.61

apped

-0.60

aved

-0.60

 dotted

-0.60

ked

-0.60

POSITIVE LOGITS

 Edition

1.15

edition

0.80

 Reborn

0.79

erity

0.77

smanship

0.73

bringer

0.72

ctl

0.71

osaurus

0.71

letes

0.70

wards

0.69

Activations Density 0.009%

terms related to special or limited editions of products

references to various editions of products or media

No Comments

No Known Activations