INDEX

Explanations

words related to questionable actions or characteristics

oai_token-act-pair · gpt-3.5-turbo

phrases that express skepticism or doubt about integrity or quality

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 0-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.0.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.0.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

itect

-0.77

ingle

-0.71

aman

-0.67

rients

-0.66

 unite

-0.64

Cube

-0.63

 Furn

-0.63

ette

-0.63

Feed

-0.62

\">

-0.62

POSITIVE LOGITS

 questionable

3.33

 dubious

3.16

 doubtful

2.17

 shaky

1.79

 shady

1.68

 murky

1.60

 controversial

1.47

 bogus

1.44

 uncertain

1.43

 suspicious

1.42

Activations Density 0.028%

words related to questionable actions or characteristics

phrases that express skepticism or doubt about integrity or quality

No Comments

No Known Activations