INDEX

Explanations

the word "are" at a high activation level

oai_token-act-pair · gpt-3.5-turbo

phrases indicating probability or likelihood

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

azel

-0.79

chall

-0.78

 rele

-0.69

kind

-0.68

 streng

-0.67

inav

-0.67

Els

-0.67

coh

-0.66

lik

-0.65

omics

-0.65

POSITIVE LOGITS

 elapsed

0.82

aneously

0.81

 expires

0.72

 WHEN

0.68

 hesitation

0.67

PST

0.67

 separating

0.66

*/(

0.66

PDT

0.66

CST

0.66

Activations Density 0.308%

the word "are" at a high activation level

phrases indicating probability or likelihood

No Comments

No Known Activations