INDEX

Explanations

phrases related to strong personal convictions and promises

oai_token-act-pair · gpt-3.5-turbo

expressions of permanence or unchangeable circumstances

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

IFrame

Link

Not in Any Lists

No Comments

Negative Logits

Newsletter

-0.79

-0.78

Kin

-0.74

externalActionCode

-0.69

soType

-0.68

MpServer

-0.67

DIT

-0.67

 largeDownload

-0.66

LESS

-0.65

CAR

-0.64

POSITIVE LOGITS

theless

0.99

 achieve

0.93

 underestimate

0.93

 EVER

0.90

 tolerate

0.85

 rematch

0.85

be

0.85

 admit

0.84

 become

0.84

 dream

0.84

Activations Density 0.046%

phrases related to strong personal convictions and promises

expressions of permanence or unchangeable circumstances

No Comments

No Known Activations