INDEX

Explanations

abbreviations or terms with the letter 'Q'

oai_token-act-pair · gpt-3.5-turbo

occurrences of the term "Q" followed by a number

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 2-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.2.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.2.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

éĹĺ

-0.78

 Awakens

-0.74

milo

-0.71

wagen

-0.71

 Magicka

-0.71

fulness

-0.69

 fullest

-0.69

ãĥ¼ãĥ«

-0.68

 Chaser

-0.67

-0.66

POSITIVE LOGITS

addafi

1.02

UE

0.93

atari

0.88

WER

0.88

agan

0.82

ues

0.81

aida

0.81

onga

0.80

aho

0.80

atana

0.79

Activations Density 0.014%

abbreviations or terms with the letter 'Q'

occurrences of the term "Q" followed by a number

No Comments

No Known Activations