INDEX

Explanations

objects or items located in specific positions within a larger setting, such as "at the end", "on the top" or "on the back"

oai_token-act-pair · gpt-3.5-turbo

elements and features related to visual design or layout

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

brate

-0.68

rowd

-0.65

 endure

-0.64

brates

-0.62

humans

-0.62

mith

-0.60

orses

-0.60

imore

-0.60

tests

-0.60

dogs

-0.59

POSITIVE LOGITS

 bottom

1.09

 underside

1.07

 periphery

1.04

bottom

0.91

 side

0.91

 entrance

0.91

 center

0.90

 corner

0.89

 centre

0.85

facing

0.85

Activations Density 0.350%

objects or items located in specific positions within a larger setting, such as "at the end", "on the top" or "on the back"

elements and features related to visual design or layout

No Comments

No Known Activations