INDEX

Explanations

numbers indicating a level or value of performance or achievement

oai_token-act-pair · gpt-3.5-turbo

references to scores and ratings in various contexts

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 5-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.5.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.5.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

xon

-0.75

ĸļ

-0.70

 Parenthood

-0.65

pora

-0.64

LECT

-0.61

 flirt

-0.61

 detachment

-0.61

 perspect

-0.58

lly

-0.57

agan

-0.56

POSITIVE LOGITS

card

1.27

cards

1.20

ific

1.01

keeper

0.93

keepers

0.90

 scores

0.86

aments

0.85

emi

0.84

ifications

0.84

ient

0.82

Activations Density 0.020%

numbers indicating a level or value of performance or achievement

references to scores and ratings in various contexts

No Comments

No Known Activations