INDEX

Explanations

words related to specific medical terms, particularly acronyms

oai_token-act-pair · gpt-3.5-turbo

references to medical device identifiers

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

PlotsExplanationShow Test FieldDefault Test Text

IFrame

Link

Not in Any Lists

No Comments

Negative Logits

acters

-0.74

tem

-0.73

xxxxxxxx

-0.71

flies

-0.70

=-=-=-=-=-=-=-=-

-0.69

give

-0.69

sell

-0.69

 Proposition

-0.67

=-=-=-=-

-0.67

stall

-0.67

POSITIVE LOGITS

MD

1.26

iamond

0.96

iate

0.90

 condem

0.85

inctions

0.81

ription

0.81

essage

0.81

ividual

0.80

VIDIA

0.80

rador

0.78

Activations Density 0.002%

words related to specific medical terms, particularly acronyms

references to medical device identifiers

No Comments

No Known Activations