INDEX

Explanations

locations or descriptions of spaces

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

IFrame

Link

Not in Any Lists

No Comments

Negative Logits

 Dating

-0.68

hene

-0.67

ophobia

-0.66

Merit

-0.62

 Discrimination

-0.61

:(

-0.61

quit

-0.60

THING

-0.60

 REAL

-0.60

 anymore

-0.60

POSITIVE LOGITS

 assorted

0.87

 accompan

0.82

 nearby

0.81

 respectively

0.79

 complement

0.78

cellaneous

0.78

 optionally

0.76

 accompany

0.75

 lieu

0.75

 standby

0.75

Activations Density 0.419%

locations or descriptions of spaces

No Comments

No Known Activations