INDEX

Explanations

the name "Joe" in text

oai_token-act-pair · gpt-3.5-turbo

references to the name "Joe."

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

IFrame

Link

Not in Any Lists

No Comments

Negative Logits

raints

-0.78

 glim

-0.76

hips

-0.75

HCR

-0.74

chwitz

-0.72

binary

-0.72

igators

-0.68

imental

-0.68

iment

-0.68

ylum

-0.66

POSITIVE LOGITS

 Biden

1.00

Joe

0.98

Joe

0.89

ppo

0.86

zzi

0.84

Rog

0.80

okes

0.77

athon

0.77

 Walsh

0.76

 Arpaio

0.74

Activations Density 0.008%

the name "Joe" in text

references to the name "Joe."

No Comments

No Known Activations