INDEX

Explanations

the word "probe" in various contexts

oai_token-act-pair · gpt-3.5-turbo

references to "probes" 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 @ 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

 Extrem

-0.67

Decl

-0.65

iana

-0.65

birth

-0.64

licks

-0.63

 Birth

-0.60

monton

-0.59

 compe

-0.59

Aid

-0.59

rip

-0.59

POSITIVE LOGITS

 probing

1.09

 probe

1.06

 probes

1.00

 Probe

1.00

 inquiry

0.83

 investigating

0.75

 gauge

0.75

 investigation

0.73

 questioning

0.71

 sweep

0.70

Activations Density 0.010%

the word "probe" in various contexts

references to "probes" in various contexts

No Comments

No Known Activations