INDEX

Explanations

phrases related to educational levels and training programs

oai_token-act-pair · gpt-3.5-turbo

references to hierarchical categories or classifications

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

 likely

-0.74

 bills

-0.68

Kes

-0.66

 credible

-0.66

 dams

-0.63

 vandalism

-0.63

 fascism

-0.62

 hail

-0.62

 preempt

-0.61

 seriously

-0.61

POSITIVE LOGITS

inner

4.48

outer

1.82

 inner

1.20

inning

1.18

 Inner

1.14

ender

1.08

inn

1.07

iner

1.03

ners

1.01

astern

0.99

Activations Density 0.012%

phrases related to educational levels and training programs

references to hierarchical categories or classifications

No Comments

No Known Activations