INDEX

Explanations

phrases referring to physical objects being connected or associated with something else

oai_token-act-pair · gpt-3.5-turbo

occurrences of the word "attached."

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 4-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.4.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.4.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

$$$$

-0.76

Cola

-0.74

Martin

-0.74

java

-0.70

 Kardashian

-0.68

externalActionCode

-0.67

;;;;;;;;;;;;

-0.66

Bos

-0.65

enthal

-0.65

YR

-0.65

POSITIVE LOGITS

 detach

0.94

 attached

0.87

 attach

0.87

 thereto

0.87

 satell

0.78

rity

0.76

alore

0.76

icut

0.76

ments

0.75

eret

0.72

Activations Density 0.015%

phrases referring to physical objects being connected or associated with something else

occurrences of the word "attached."

No Comments

No Known Activations