INDEX

Explanations

character sequences that appear to be technical codes or identifiers

oai_token-act-pair · gpt-3.5-turbo

special characters and specific terms related to various topics

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

ItemThumbnailImage

-0.64

 faked

-0.63

 Eisenhower

-0.62

 confidentiality

-0.62

 approximate

-0.61

 immediate

-0.61

gotten

-0.61

 Lauder

-0.60

 underlying

-0.60

 privacy

-0.59

POSITIVE LOGITS

Äĩ

0.93

ondo

0.84

merce

0.82

Ã©n

0.82

Ã¡s

0.80

quit

0.80

Äį

0.80

soDeliveryDate

0.79

uchi

0.79

Ä

0.78

Activations Density 0.564%

character sequences that appear to be technical codes or identifiers

special characters and specific terms related to various topics

No Comments

No Known Activations