INDEX

Explanations

text characters like symbols and numbers

oai_token-act-pair · gpt-3.5-turbo

tokens related to social media handles or URLs

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

PlotsExplanationShow Test FieldDefault Test Text

IFrame

Link

Not in Any Lists

No Comments

Negative Logits

 resil

-0.70

 bombard

-0.70

 arson

-0.69

 cens

-0.69

 Levant

-0.67

 behavi

-0.67

 peac

-0.66

 shorth

-0.66

 dish

-0.64

 redevelop

-0.64

POSITIVE LOGITS

fw

1.07

cd

1.06

df

1.04

zx

1.01

bc

1.01

bm

1.01

zn

0.99

wb

0.97

fb

0.97

gd

0.95

Activations Density 0.064%

text characters like symbols and numbers

tokens related to social media handles or URLs

No Comments

No Known Activations