INDEX

Explanations

words that are part of programming code

oai_token-act-pair · gpt-3.5-turbo

punctuation marks, particularly commas

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 10-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.10.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.10.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

peg

-0.61

ussy

-0.60

gag

-0.58

 upfront

-0.57

 rigs

-0.57

 isolation

-0.56

 everyday

-0.56

 undercover

-0.56

bie

-0.56

 disenfranch

-0.55

POSITIVE LOGITS

actionDate

0.97

pmwiki

0.78

 Accessed

0.76

,,,,,,,,

0.75

ĸļ

0.75

align

0.75

soDeliveryDate

0.75

pp

0.74

published

0.74

channelAvailability

0.73

Activations Density 0.058%

words that are part of programming code

punctuation marks, particularly commas

No Comments

No Known Activations