INDEX

Explanations

phrases related to specific terms and abbreviations, potentially related to technology, law, or business

oai_token-act-pair · gpt-3.5-turbo

parentheses and brackets in the text

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 8-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.8.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.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

 cous

-0.73

 reefs

-0.72

othal

-0.69

 dred

-0.69

 skew

-0.68

osaurus

-0.67

lav

-0.67

exh

-0.66

 squid

-0.66

 fins

-0.65

POSITIVE LOGITS

Independent

1.01

meaning

0.99

Elect

0.96

short

0.91

pron

0.88

independent

0.87

sic

0.87

Community

0.85

Universal

0.84

Simple

0.84

Activations Density 0.092%

phrases related to specific terms and abbreviations, potentially related to technology, law, or business

parentheses and brackets in the text

No Comments

No Known Activations