INDEX

Explanations

descriptions of actions or suggestions related to helping or aiding others

oai_token-act-pair · gpt-3.5-turbo

pronouns and actions related to potential or capability

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 10-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.10.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.10.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

 sqor

-0.78

 millenn

-0.65

 partName

-0.64

 Cosponsors

-0.62

 attm

-0.59

 Arri

-0.58

 fraught

-0.58

 treacher

-0.58

 lately

-0.57

Nay

-0.57

POSITIVE LOGITS

can

1.13

 wouldn

1.03

 could

1.02

 wont

1.02

'll

0.91

 dont

0.91

won

0.85

sylv

0.84

 avoids

0.80

 maxim

0.79

Activations Density 0.154%

descriptions of actions or suggestions related to helping or aiding others

pronouns and actions related to potential or capability

No Comments

No Known Activations