INDEX

Explanations

web-related technical code snippets

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

fulness

-0.73

fully

-0.72

wards

-0.72

ffield

-0.70

cover

-0.69

 unavoid

-0.69

chal

-0.68

 sheer

-0.66

 shocks

-0.65

gger

-0.65

POSITIVE LOGITS

 ..........

1.38

VERTISEMENT

0.93

Â¯Â¯Â¯Â¯Â¯Â¯Â¯Â¯

0.79

 Utilities

0.78

Skip

0.74

 Albuquerque

0.73

 FILE

0.73

 COUNTY

0.73

TION

0.70

inances

0.69

Activations Density 0.020%

web-related technical code snippets

No Comments

No Known Activations