INDEX

Explanations

words related to legal terms or expressions

oai_token-act-pair · gpt-3.5-turbo

sequences of letters or abbreviations

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

DragonMagazine

-0.90

glers

-0.76

Magikarp

-0.76

ÄŁ

-0.68

senal

-0.67

chet

-0.66

 Clover

-0.64

clinton

-0.64

auga

-0.64

 Turkey

-0.63

POSITIVE LOGITS

inct

0.89

ossession

0.85

erest

0.84

ributed

0.83

ussions

0.82

erves

0.82

resso

0.81

isive

0.80

acement

0.79

unction

0.78

Activations Density 0.113%

words related to legal terms or expressions

sequences of letters or abbreviations

No Comments

No Known Activations