INDEX

Explanations

numerical rankings or positions

oai_token-act-pair · gpt-3.5-turbo

references to "top" rankings or positions in various contexts

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

PlotsExplanationShow Test FieldDefault Test Text

IFrame

Link

Not in Any Lists

No Comments

Negative Logits

 Manual

-0.71

 Territories

-0.68

 Neon

-0.66

 Gaul

-0.64

 Griffith

-0.64

arij

-0.63

 Lover

-0.63

 Anth

-0.62

iencies

-0.62

 Arri

-0.62

POSITIVE LOGITS

most

1.04

 tier

1.03

top

0.94

top

0.90

ographical

0.87

 notch

0.86

 scorer

0.86

deck

0.84

TOP

0.82

Top

0.81

Activations Density 0.033%

numerical rankings or positions

references to "top" rankings or positions in various contexts

No Comments

No Known Activations