INDEX

Explanations

names of places or events

oai_token-act-pair · gpt-3.5-turbo

references to specific events or titles related to pop culture or media

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

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

theless

-0.68

 admitting

-0.65

 related

-0.64

 oneself

-0.61

 mechanically

-0.61

 approving

-0.60

 reporting

-0.60

 offline

-0.59

 centrally

-0.58

 unfavorable

-0.58

POSITIVE LOGITS

agogue

0.90

isine

0.88

tones

0.86

orum

0.85

astery

0.84

 Orchestra

0.83

istries

0.82

 Remix

0.82

inery

0.80

adas

0.79

Activations Density 0.507%

names of places or events

references to specific events or titles related to pop culture or media

No Comments

No Known Activations