INDEX

Explanations

phrases related to medical emergencies and hospital treatments

oai_token-act-pair · gpt-3.5-turbo

mentions of hospitals and medical treatment in relation to victims

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 11-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.11.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.11.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

yip

-0.79

 unlaw

-0.62

 penet

-0.61

dust

-0.61

hire

-0.60

ces

-0.60

crop

-0.60

idable

-0.59

bill

-0.59

 eaves

-0.59

POSITIVE LOGITS

 recovering

0.94

 suffering

0.88

 undergoing

0.85

 hospitalized

0.83

 Hospital

0.81

 hospital

0.80

ospital

0.79

 doctors

0.77

 intensive

0.75

 stabilized

0.75

Activations Density 0.117%

phrases related to medical emergencies and hospital treatments

mentions of hospitals and medical treatment in relation to victims

No Comments

No Known Activations