INDEX

Explanations

items listed or mentioned sequentially

oai_token-act-pair · gpt-3.5-turbo

items or concepts related to lists, categories, and organizational structures

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

IFrame

Link

Not in Any Lists

No Comments

Negative Logits

 arresting

-0.63

ãĥ©ãĥ³

-0.62

akin

-0.61

 inflicting

-0.58

ÃŃn

-0.57

 imposing

-0.57

iaz

-0.57

angering

-0.57

 foreign

-0.56

 fearing

-0.54

POSITIVE LOGITS

 varies

1.18

 revolves

1.14

 consists

1.05

 boils

1.03

 depends

1.01

 tends

0.98

 differs

0.96

 reminds

0.96

 sucks

0.94

:)

0.93

Activations Density 0.460%

items listed or mentioned sequentially

items or concepts related to lists, categories, and organizational structures

No Comments

No Known Activations