INDEX

Explanations

words related to causing strong emotional or physical reactions, particularly through intense actions or events

oai_token-act-pair · gpt-3.5-turbo

phrases that contain the word "up."

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

hari

-0.72

 pierced

-0.66

NetMessage

-0.66

cheon

-0.64

 Closing

-0.62

Obj

-0.62

Bei

-0.60

Gi

-0.60

Ree

-0.59

_>

-0.59

POSITIVE LOGITS

xual

0.76

river

0.70

amac

0.70

grading

0.67

 fists

0.66

draft

0.66

ãĤ±

0.65

ãĤ¤ãĥĪ

0.65

orsche

0.64

 havoc

0.61

Activations Density 0.040%

words related to causing strong emotional or physical reactions, particularly through intense actions or events

phrases that contain the word "up."

No Comments

No Known Activations