INDEX

Explanations

words related to shortcuts or quick ways to achieve something

oai_token-act-pair · gpt-3.5-turbo

references to shortcuts 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

ordon

-0.87

ighth

-0.86

iaries

-0.82

olicited

-0.82

itor

-0.81

uo

-0.81

ock

-0.80

erva

-0.80

zona

-0.80

orio

-0.79

POSITIVE LOGITS

 shortcut

0.85

 shortcuts

0.82

 towed

0.69

 snipp

0.68

 annexed

0.67

ABE

0.67

 hint

0.66

step

0.66

 procedure

0.64

 method

0.64

Activations Density 0.016%

words related to shortcuts or quick ways to achieve something

references to shortcuts in various contexts

No Comments

No Known Activations