INDEX

Explanations

statements involving actions or events happening abruptly or unexpectedly

oai_token-act-pair · gpt-3.5-turbo

statements of intent or expectation

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

obar

-0.73

 subsidized

-0.69

Ń·

-0.69

arthed

-0.68

rosis

-0.67

ĸļ

-0.67

¥µ

-0.66

 Cerberus

-0.65

ibrary

-0.64

utical

-0.64

POSITIVE LOGITS

Changes

0.92

Looking

0.85

Calling

0.84

However

0.83

Resp

0.82

Instead

0.81

Companies

0.81

Making

0.80

Import

0.80

Whether

0.80

Activations Density 0.189%

statements involving actions or events happening abruptly or unexpectedly

statements of intent or expectation

No Comments

No Known Activations