INDEX

Explanations

references to explosive events or related situations

oai_token-act-pair · gpt-3.5-turbo

references to explosions or blasts

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

ccording

-0.89

pires

-0.84

gemony

-0.81

prison

-0.77

 compr

-0.76

Serial

-0.75

phis

-0.75

Decre

-0.75

guyen

-0.75

Development

-0.70

POSITIVE LOGITS

 blast

1.35

 blasts

1.13

 furnace

0.92

blast

0.89

 furn

0.85

waves

0.78

ocy

0.78

 bursting

0.77

 showers

0.77

astically

0.77

Activations Density 0.010%

references to explosive events or related situations

references to explosions or blasts

No Comments

No Known Activations