INDEX

Explanations

programming code syntax and technology-related terms

oai_token-act-pair · gpt-3.5-turbo

the word "and" in various contexts

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 3-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.3.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.3.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

bub

-0.83

Deal

-0.75

natureconservancy

-0.66

cause

-0.65

arthed

-0.65

amer

-0.64

nw

-0.64

Wisconsin

-0.63

atten

-0.63

POSITIVE LOGITS

 therefore

1.05

 then

0.99

rogen

0.96

 consequently

0.96

rogens

0.92

hra

0.88

romeda

0.88

 thus

0.85

alus

0.84

 thence

0.84

Activations Density 0.513%

programming code syntax and technology-related terms

the word "and" in various contexts

No Comments

No Known Activations