INDEX

Explanations

phrases or clauses that express an opinion or claim

oai_token-act-pair · gpt-3.5-turbo

instances of the word "said" followed by punctuation, indicating speech or reported dialogue

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 3-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.3.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.3.hook_resid_pre

Architecture

standard

Context Size

128

Dataset

Skylion007/openwebtext

Hook Point Layer

Activation Function

relu

Embeds

IFrame

Link

Not in Any Lists

No Comments

Negative Logits

gate

-0.73

aval

-0.65

olves

-0.63

swick

-0.60

interstitial

-0.59

hei

-0.58

 cess

-0.57

 furnace

-0.55

ometers

-0.55

herent

-0.54

POSITIVE LOGITS

 citing

0.88

 although

0.86

 albeit

0.82

 quoting

0.79

 "âĢ¦

0.75

 namely

0.72

sir

0.71

"'

0.70

 adding

0.68

 anecd

0.68

Activations Density 0.451%

phrases or clauses that express an opinion or claim

instances of the word "said" followed by punctuation, indicating speech or reported dialogue

No Comments

No Known Activations