INDEX

Explanations

names followed by the action of telling or informing

oai_token-act-pair · gpt-3.5-turbo

instances of the word "told" in various contexts

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 10-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.10.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.10.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

isol

-0.76

 ILCS

-0.73

imposed

-0.68

artifacts

-0.68

adesh

-0.66

fet

-0.66

ãĥİ

-0.65

aband

-0.65

execute

-0.64

ciplinary

-0.64

POSITIVE LOGITS

 reporters

1.27

 HuffPost

1.13

 BuzzFeed

1.11

CNN

1.09

 interviewer

1.08

me

1.06

 CNBC

1.02

 VICE

1.01

 Politico

0.98

 MSNBC

0.98

Activations Density 0.054%

names followed by the action of telling or informing

instances of the word "told" in various contexts

No Comments

No Known Activations