INDEX

Explanations

phrases related to success or performance evaluation

oai_token-act-pair · gpt-3.5-turbo

phrases related to someone's historical performance or credibility

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 5-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.5.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.5.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

essel

-0.87

edi

-0.78

sub

-0.75

ishable

-0.73

 uncond

-0.73

hap

-0.73

andr

-0.71

tein

-0.70

amaz

-0.70

adin

-0.70

POSITIVE LOGITS

 record

0.81

 breaker

0.79

record

0.68

 recorder

0.68

 Record

0.67

 holder

0.67

olester

0.66

 achievement

0.65

 Choice

0.64

 achievements

0.63

Activations Density 0.035%

phrases related to success or performance evaluation

phrases related to someone's historical performance or credibility

No Comments

No Known Activations