INDEX

Explanations

phrases related to things that are often neglected or not given enough attention

oai_token-act-pair · gpt-3.5-turbo

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

usalem

-0.70

 Mood

-0.61

ylene

-0.58

hedral

-0.58

idine

-0.57

lua

-0.57

sei

-0.56

 Aires

-0.56

ergy

-0.55

FI

-0.55

POSITIVE LOGITS

 anymore

0.92

 altogether

0.91

 nowadays

0.87

 amidst

0.84

 lest

0.84

by

0.80

 until

0.78

 unless

0.78

 amid

0.75

 amongst

0.74

Activations Density 0.195%

phrases related to things that are often neglected or not given enough attention

No Comments

No Known Activations