INDEX

Explanations

mentions of medical emergencies, medical conditions, and healthcare-related situations

oai_token-act-pair · gpt-3.5-turbo Triggered by @bot

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 8-res_fs768-jb

Configuration

jbloom/GPT2-Small-Feature-Splitting-Experiment-Layer-8/blocks.8.hook_resid_pre_768

Prompts (Dashboard)

24,576 prompts, 128 tokens each

Dataset (Dashboard)

Skylion007/openwebtext

Features

768

Data Type

torch.float32

Hook Point

blocks.8.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

Publisher

-0.79

*/(

-0.77

 Strongh

-0.71

izont

-0.68

bour

-0.66

xtap

-0.66

 Unch

-0.66

 rand

-0.65

ickr

-0.65

 Grayson

-0.64

POSITIVE LOGITS

 patients

1.28

 diagnoses

1.18

 patient

1.17

 Patients

1.16

 care

1.14

 doctor

1.12

 diagnosis

1.11

 surgeries

1.11

 Patient

1.10

 doctors

1.10

Activations Density 4.021%

mentions of medical emergencies, medical conditions, and healthcare-related situations

No Comments

No Known Activations