INDEX

Explanations

words related to breaking or exceeding limits or rules

oai_token-act-pair · gpt-3.5-turbo

instances of the abbreviation "AF" or variations of "af."

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

 Fargo

-0.72

 Patriarch

-0.69

 Mothers

-0.66

DOWN

-0.64

 disinfect

-0.64

 Mour

-0.63

PASS

-0.63

MER

-0.62

 expectancy

-0.61

offer

-0.61

POSITIVE LOGITS

rican

1.26

rica

1.16

icion

1.09

ghan

0.96

riad

0.96

avorite

0.95

athom

0.93

riend

0.90

eties

0.89

eatures

0.87

Activations Density 0.010%

words related to breaking or exceeding limits or rules

instances of the abbreviation "AF" or variations of "af."

No Comments

No Known Activations