INDEX

Explanations

mentions of medical conditions and hospitals in news or descriptions of incidents

oai_token-act-pair · gpt-3.5-turbo

references to hospitalizations and medical conditions following incidents or injuries

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 10-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.10.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.10.hook_resid_pre

Architecture

standard

Context Size

128

Dataset

Skylion007/openwebtext

Hook Point Layer

Activation Function

relu

Embeds

PlotsExplanationShow Test FieldDefault Test Text

IFrame

Link

Not in Any Lists

No Comments

Negative Logits

repeat

-0.64

 Strongh

-0.64

estamp

-0.64

idable

-0.63

skill

-0.62

wcsstore

-0.61

 unlaw

-0.61

pas

-0.59

 Millenn

-0.58

 neigh

-0.58

POSITIVE LOGITS

ised

0.99

ization

0.91

 hospitalized

0.87

 recovering

0.85

 hospital

0.83

ized

0.81

ospital

0.79

 healed

0.79

itals

0.78

 doctors

0.78

Activations Density 0.035%

mentions of medical conditions and hospitals in news or descriptions of incidents

references to hospitalizations and medical conditions following incidents or injuries

No Comments

No Known Activations