INDEX

Explanations

words related to checking off items on a list or receiving validation

oai_token-act-pair · gpt-3.5-turbo

references to the concept of ticking, particularly in relation to emotions, reactions, or mechanisms

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 1-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.1.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.1.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

ocally

-0.75

éĹĺ

-0.66

ht

-0.65

 wills

-0.62

 ILCS

-0.62

ufact

-0.62

 Kore

-0.60

ãĥ´ãĤ¡

-0.59

perty

-0.59

images

-0.59

POSITIVE LOGITS

 tick

1.19

tick

1.10

 ticks

1.08

 Tick

0.96

lers

0.81

ering

0.81

len

0.80

 Clicker

0.79

ety

0.75

lish

0.75

Activations Density 0.003%

words related to checking off items on a list or receiving validation

references to the concept of ticking, particularly in relation to emotions, reactions, or mechanisms

No Comments

No Known Activations