INDEX

Explanations

references to Twitter tweets and news articles

oai_token-act-pair · gpt-3.5-turbo

punctuation and special characters

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

 apprentices

-0.73

 favour

-0.69

 relocation

-0.69

 disciplines

-0.69

enary

-0.67

 furn

-0.66

emouth

-0.66

 teleport

-0.65

 specialization

-0.65

 spaced

-0.64

POSITIVE LOGITS

Screenshot

1.13

Actor

1.02

Twitter

1.01

Attempts

0.95

CNN

0.95

Twe

0.95

Comment

0.94

Earlier

0.94

Shortly

0.94

Around

0.94

Activations Density 0.516%

references to Twitter tweets and news articles

punctuation and special characters

No Comments

No Known Activations