INDEX

Explanations

words related to the conclusion or termination of events or periods

oai_token-act-pair · gpt-3.5-turbo

instances of the word "ended."

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

esis

-0.78

itsch

-0.74

alty

-0.74

hee

-0.73

king

-0.73

icipated

-0.73

uni

-0.70

uana

-0.68

arden

-0.68

atana

-0.67

POSITIVE LOGITS

angering

0.97

 tragically

0.90

owment

0.87

 abruptly

0.87

 prematurely

0.84

ocrin

0.80

 sidel

0.69

ocrine

0.68

ocument

0.68

Ń·

0.68

Activations Density 0.025%

words related to the conclusion or termination of events or periods

instances of the word "ended."

No Comments

No Known Activations