INDEX

Explanations

references to seating arrangements

oai_token-act-pair · gpt-3.5-turbo

references to "seat" in various contexts

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 0-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.0.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.0.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

 dotted

-0.77

 green

-0.71

 recomb

-0.68

Lew

-0.67

 caution

-0.66

 greed

-0.66

rem

-0.65

 cycl

-0.65

 Frazier

-0.65

bio

-0.65

POSITIVE LOGITS

seat

4.17

 seat

1.72

 Seat

1.66

 seats

1.49

vote

1.13

slot

1.13

ride

1.12

lane

1.07

rider

1.01

deck

1.01

Activations Density 0.021%

references to seating arrangements

references to "seat" in various contexts

No Comments

No Known Activations