INDEX

Explanations

terms related to electrical engineering variables and symbols

oai_token-act-pair · gpt-3.5-turbo

punctuations and conjunctions in lists or structured data

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

IFrame

Link

Not in Any Lists

No Comments

Negative Logits

elaide

-0.70

hement

-0.69

thinkable

-0.64

ipment

-0.63

heed

-0.60

 convictions

-0.59

ipped

-0.58

 afore

-0.58

scrib

-0.58

 assassination

-0.57

POSITIVE LOGITS

 Î²

0.97

0.96

XL

0.95

Î

0.95

Ïī

0.89

VG

0.89

0.88

YC

0.87

XI

0.87

LF

0.83

Activations Density 0.103%

terms related to electrical engineering variables and symbols

punctuations and conjunctions in lists or structured data

No Comments

No Known Activations