INDEX

Explanations

phrases related to official information or regulations

oai_token-act-pair · gpt-3.5-turbo

critical safety information or warnings

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 1-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.1.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.1.hook_resid_pre

Architecture

standard

Context Size

128

Dataset

Skylion007/openwebtext

Hook Point Layer

Activation Function

relu

Embeds

PlotsExplanationShow Test FieldDefault Test Text

IFrame

Link

Not in Any Lists

No Comments

Negative Logits

Streamer

-0.76

¥

-0.73

VK

-0.71

Ħ

-0.68

¼

-0.67

¡

-0.66

EMOTE

-0.65

sys

-0.65

Allows

-0.65

[+]

-0.64

POSITIVE LOGITS

arde

0.79

atown

0.75

 nodd

0.73

 neighb

0.72

 canv

0.70

 polic

0.70

 reluct

0.70

 conduc

0.67

 ende

0.67

ourage

0.67

Activations Density 0.000%

phrases related to official information or regulations

critical safety information or warnings

No Comments

No Known Activations