INDEX

Explanations

phrases related to things that have been demonstrated or confirmed as factual or effective

oai_token-act-pair · gpt-3.5-turbo

phrases emphasizing evidence of reliability or established success

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

adish

-0.73

 newsletters

-0.68

ifle

-0.67

idays

-0.66

 sshd

-0.64

iewicz

-0.64

letal

-0.64

umbn

-0.63

hhh

-0.63

eeper

-0.63

POSITIVE LOGITS

ãĥ¼ãĥĨ

0.91

 proven

0.87

iary

0.84

uable

0.81

\\\\\\\\

0.79

Ø³

0.78

ance

0.77

Ø¯

0.73

ãĤ¤ãĥĪ

0.72

 debunked

0.71

Activations Density 0.019%

phrases related to things that have been demonstrated or confirmed as factual or effective

phrases emphasizing evidence of reliability or established success

No Comments

No Known Activations