INDEX

Explanations

mentions of the word "Channel"

oai_token-act-pair · gpt-4-turbo

mentions of specific television channels and news programs

oai_token-act-pair · gpt-3.5-turbo

mentions of television channels

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

earable

-0.70

olding

-0.70

arcity

-0.67

ever

-0.67

 sund

-0.64

Rao

-0.64

 Oilers

-0.64

sson

-0.64

nexpected

-0.62

ĻĤ

-0.61

POSITIVE LOGITS

Channel

1.13

 Channel

1.11

 channel

0.88

 Divinity

0.86

Mask

0.86

 Tunnel

0.83

 Islands

0.82

annels

0.79

channel

0.78

icut

0.77

Activations Density 0.008%

mentions of the word "Channel"

mentions of specific television channels and news programs

mentions of television channels

No Comments

No Known Activations