INDEX

Explanations

phrases related to general concepts or ideas

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

 unequ

-0.70

ãĥĺ

-0.63

 ILCS

-0.60

ochet

-0.60

entimes

-0.59

attering

-0.59

Neg

-0.58

ategor

-0.56

thood

-0.55

fashion

-0.55

POSITIVE LOGITS

 derives

0.96

 revolves

0.95

 behind

0.93

 consists

0.91

 stems

0.86

 rationale

0.85

of

0.84

 consisted

0.81

 refers

0.81

 differs

0.80

Activations Density 0.242%

phrases related to general concepts or ideas

No Comments

No Known Activations