INDEX

Explanations

words related to interactions and engagement with an audience

oai_token-act-pair · gpt-3.5-turbo

phrases indicating audience acceptance or engagement

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 11-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.11.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.11.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

WithNo

-0.80

ĨĴ

-0.78

ELD

-0.70

ãĥĺãĥ©

-0.69

ãĤ¬

-0.67

ãĥĩãĤ£

-0.66

ãĤ´

-0.65

itary

-0.65

ãĤ¦ãĤ¹

-0.63

ARB

-0.62

POSITIVE LOGITS

 approve

1.25

 respond

1.16

 react

1.15

 disapprove

1.13

 agree

1.09

 reciproc

1.09

 laugh

1.07

 flock

1.05

 notice

1.04

 balk

1.02

Activations Density 0.393%

words related to interactions and engagement with an audience

phrases indicating audience acceptance or engagement

No Comments

No Known Activations