INDEX

Explanations

recent or past events or statements

oai_token-act-pair · gpt-3.5-turbo

repeated references to the word "ago," indicating discussions about time and past events

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 3-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.3.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.3.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

 mathemat

-0.76

 plaus

-0.74

 uniqueness

-0.73

 suspic

-0.71

 simultaneous

-0.69

 glim

-0.69

 illumination

-0.69

 tremend

-0.68

sheet

-0.68

 similarity

-0.68

POSITIVE LOGITS

vernment

1.27

zzi

1.08

ago

1.05

opa

0.97

edia

0.96

zzo

0.94

onga

0.92

onte

0.91

etta

0.88

allo

0.87

Activations Density 0.007%

recent or past events or statements

repeated references to the word "ago," indicating discussions about time and past events

No Comments

No Known Activations