INDEX

Explanations

phrases indicating certainty or correctness

oai_token-act-pair · gpt-3.5-turbo

statements asserting correctness or verification of information

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 11-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.11.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.11.hook_resid_pre

Architecture

standard

Context Size

128

Dataset

Skylion007/openwebtext

Hook Point Layer

Activation Function

relu

Embeds

IFrame

Link

Not in Any Lists

No Comments

Negative Logits

icipated

-0.76

venant

-0.73

ulton

-0.72

estine

-0.71

clerosis

-0.70

berus

-0.68

igslist

-0.67

letal

-0.66

TOD

-0.65

 Installation

-0.64

POSITIVE LOGITS

 referring

1.49

 correct

1.44

 wrong

1.41

 quoting

1.32

 mistaken

1.25

 kidding

1.24

 incorrect

1.22

 arguing

1.21

 talking

1.20

 joking

1.19

Activations Density 0.240%

phrases indicating certainty or correctness

statements asserting correctness or verification of information

No Comments

No Known Activations