INDEX

Explanations

words related to software, programming, and technical instructions

oai_token-act-pair · gpt-3.5-turbo

punctuation, specifically commas

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 4-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.4.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.4.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

¬¼

-0.73

ahu

-0.67

uscript

-0.64

 crisis

-0.63

untarily

-0.63

 advis

-0.61

irie

-0.60

enance

-0.60

ery

-0.59

 Yemen

-0.58

POSITIVE LOGITS

etc

1.21

 haha

1.17

 namely

1.09

 preferably

1.06

aka

1.05

ie

1.02

 regardless

0.96

 albeit

0.95

etc

0.95

 whereas

0.94

Activations Density 1.101%

words related to software, programming, and technical instructions

punctuation, specifically commas

No Comments

No Known Activations