INDEX

Explanations

phrases indicating negation or opposition

oai_token-act-pair · gpt-3.5-turbo

punctuation and indicators of non-affirmative or negative statements

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

zel

-0.74

reet

-0.73

haul

-0.72

ilated

-0.66

amiliar

-0.65

osite

-0.64

eworks

-0.64

 nuts

-0.60

 redes

-0.60

estinal

-0.60

POSITIVE LOGITS

nor

1.87

 anymore

1.51

yet

1.37

Nor

1.11

nor

1.10

unless

0.96

 Instead

0.96

 unless

0.94

 Nonetheless

0.90

 whatsoever

0.87

Activations Density 0.940%

phrases indicating negation or opposition

punctuation and indicators of non-affirmative or negative statements

No Comments

No Known Activations