INDEX

Explanations

names of television channels and news programs

oai_token-act-pair · gpt-3.5-turbo

news outlets and their reporting activities

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 11-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.11.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.11.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

 ãĤµãĥ¼ãĥĨãĤ£ãĥ¯ãĥ³

-0.78

sed

-0.72

 Dalai

-0.71

Â¯

-0.67

etting

-0.66

 Lean

-0.64

ause

-0.63

FML

-0.63

 Hawking

-0.63

otrop

-0.63

POSITIVE LOGITS

ewitness

1.09

 reporter

1.02

radio

0.97

Channel

0.96

 Investig

0.95

 crews

0.95

TV

0.94

 Investigator

0.94

FOX

0.93

 affiliate

0.92

Activations Density 0.104%

names of television channels and news programs

news outlets and their reporting activities

No Comments

No Known Activations