INDEX

Explanations

references to space probes

oai_token-act-pair · gpt-3.5-turbo

references to investigative or exploratory inquiries

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 4-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.4.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.4.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

 Extrem

-0.70

 compe

-0.66

Decl

-0.66

iana

-0.64

 reconc

-0.64

AMES

-0.62

licks

-0.62

relative

-0.61

arus

-0.61

arin

-0.60

POSITIVE LOGITS

 probe

1.07

 probing

1.04

 Probe

0.99

 probes

0.97

 inquiry

0.89

 gauge

0.81

agher

0.79

 investigation

0.75

 investigating

0.72

ctr

0.72

Activations Density 0.007%

references to space probes

references to investigative or exploratory inquiries

No Comments

No Known Activations