INDEX

Explanations

phrases related to a specific product or competition

oai_token-act-pair · gpt-3.5-turbo

references to the "Pro" series, indicating a focus on professional-level content or topics

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 5-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.5.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.5.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

ãĤ¼ãĤ¦ãĤ¹

-0.84

 halls

-0.79

owship

-0.78

 straw

-0.72

 Dise

-0.70

 Cornell

-0.69

 spears

-0.66

ibrary

-0.65

 needles

-0.65

 Carrie

-0.64

POSITIVE LOGITS

digy

1.52

posal

1.23

gression

1.19

secution

1.19

verbs

1.19

blems

1.12

ceed

1.10

poses

1.09

clamation

1.08

pose

1.07

Activations Density 0.020%

phrases related to a specific product or competition

references to the "Pro" series, indicating a focus on professional-level content or topics

No Comments

No Known Activations