INDEX

Explanations

news channels and related information

oai_token-act-pair · gpt-3.5-turbo

references to news outlets and television broadcasts

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

 Â¥

-0.81

wikipedia

-0.81

 flush

-0.75

 takeoff

-0.69

ularity

-0.68

 bloom

-0.67

 melting

-0.67

hes

-0.66

 packing

-0.63

estamp

-0.62

POSITIVE LOGITS

 Investig

1.13

 Investigative

0.94

 Productions

0.93

 News

0.91

 Networks

0.89

NEWS

0.88

 Television

0.87

CTV

0.86

 Action

0.84

TV

0.83

Activations Density 0.078%

news channels and related information

references to news outlets and television broadcasts

No Comments

No Known Activations