INDEX

Explanations

words related to winning or being victorious

oai_token-act-pair · gpt-3.5-turbo

instances of the word "won" and its variations

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 1-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.1.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.1.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

ħĭ

-0.75

OTOS

-0.72

 flares

-0.65

ancies

-0.64

 clad

-0.62

 hypers

-0.61

 bonding

-0.61

 privileged

-0.59

 supervised

-0.59

 PROT

-0.59

POSITIVE LOGITS

Won

1.19

't

1.17

won

0.99

ners

0.90

cheon

0.90

hof

0.83

atan

0.82

kish

0.80

itive

0.79

geon

0.78

Activations Density 0.003%

words related to winning or being victorious

instances of the word "won" and its variations

No Comments

No Known Activations