INDEX

Explanations

words related to scientific concepts and technical details

oai_token-act-pair · gpt-3.5-turbo

terms related to explanations and underlying principles

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

udi

-0.61

IFIED

-0.61

entimes

-0.61

irrel

-0.60

alone

-0.60

inka

-0.59

Ń·

-0.58

Ħ¢

-0.58

culosis

-0.57

irgin

-0.57

POSITIVE LOGITS

 behind

1.49

behind

1.24

 involved

1.10

 underpin

1.07

 Behind

1.02

 surrounding

1.01

 workings

1.00

 underlying

0.92

of

0.87

 beneath

0.85

Activations Density 0.323%

words related to scientific concepts and technical details

terms related to explanations and underlying principles

No Comments

No Known Activations