INDEX

Explanations

words related to evoking emotions or reactions, whether positive or negative

oai_token-act-pair · gpt-3.5-turbo

words related to expressions of humor or jokes

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 4-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.4.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.4.hook_resid_pre

Architecture

standard

Context Size

128

Dataset

Skylion007/openwebtext

Hook Point Layer

Activation Function

relu

Embeds

IFrame

Link

Not in Any Lists

No Comments

Negative Logits

inance

-0.72

elsen

-0.67

ainer

-0.62

 overrun

-0.62

essors

-0.62

Lub

-0.61

 Canad

-0.59

ħ

-0.59

 Finance

-0.57

 Lama

-0.57

POSITIVE LOGITS

oking

1.03

creen

0.95

oked

0.90

 sidx

0.82

otine

0.81

=#

0.80

rue

0.77

oke

0.75

hire

0.74

EAR

0.73

Activations Density 0.019%

words related to evoking emotions or reactions, whether positive or negative

words related to expressions of humor or jokes

No Comments

No Known Activations