INDEX

Explanations

technical terms and product names related to technology and gaming

oai_token-act-pair · gpt-3.5-turbo

names of people and titles of their work or positions

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

IFrame

Link

Not in Any Lists

No Comments

Negative Logits

 unconstitutional

-0.82

 aspirin

-0.75

 constitutionally

-0.74

 constitutional

-0.74

 repeal

-0.73

 prosecutions

-0.73

 nonviolent

-0.72

 taxpayer

-0.72

 corro

-0.71

 cler

-0.70

POSITIVE LOGITS

 teased

1.35

 commented

1.19

 hinted

1.16

 stated

1.14

 divul

1.14

 reminis

1.11

 stating

1.11

 clarified

1.11

 revealed

1.09

 elaborated

1.07

Activations Density 0.612%

technical terms and product names related to technology and gaming

names of people and titles of their work or positions

No Comments

No Known Activations