INDEX

Explanations

instances where investigations or examinations are being conducted

oai_token-act-pair · gpt-3.5-turbo

occurrences of the word "probe" 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 @ 1-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.1.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.1.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

-0.68

Present

-0.63

 compe

-0.62

 Hurricanes

-0.62

awed

-0.62

reditary

-0.61

perm

-0.61

dec

-0.60

Production

-0.60

Dub

-0.59

POSITIVE LOGITS

 probe

1.36

 probes

1.25

 Probe

1.13

 probing

1.12

 inquiry

0.86

 investigation

0.84

 Inquiry

0.74

 investigating

0.74

 dred

0.70

igate

0.69

Activations Density 0.010%

instances where investigations or examinations are being conducted

occurrences of the word "probe" in various contexts

No Comments

No Known Activations