INDEX

Explanations

phrases related to medical procedures and care, particularly intensive care

oai_token-act-pair · gpt-3.5-turbo

terms related to intensive care and high-demand situations

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

flies

-0.67

 Prophet

-0.65

allows

-0.63

 Panc

-0.63

dule

-0.63

âĢ¢âĢ¢âĢ¢âĢ¢

-0.62

DPR

-0.62

Ĭ±

-0.62

 Rebell

-0.61

 Gaddafi

-0.59

POSITIVE LOGITS

 intensive

1.23

intensive

1.07

mble

0.93

icult

0.89

ly

0.88

inently

0.85

rontal

0.84

irtual

0.84

ensive

0.82

entimes

0.79

Activations Density 0.006%

phrases related to medical procedures and care, particularly intensive care

terms related to intensive care and high-demand situations

No Comments

No Known Activations