INDEX

Explanations

instances where something is incorrect or not as expected

oai_token-act-pair · gpt-3.5-turbo

instances of the word "wrong."

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 4-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.4.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.4.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

doms

-0.79

anism

-0.76

gat

-0.71

md

-0.70

dom

-0.70

lua

-0.69

ivism

-0.69

lishes

-0.68

ttes

-0.67

zen

-0.66

POSITIVE LOGITS

 thing

0.94

 side

0.87

 amount

0.86

icter

0.81

 piece

0.81

 dose

0.79

 kind

0.79

 person

0.79

 combination

0.78

guy

0.78

Activations Density 0.031%

instances where something is incorrect or not as expected

instances of the word "wrong."

No Comments

No Known Activations