INDEX

Explanations

percentages or numerical values presented in a specific format

oai_token-act-pair · gpt-3.5-turbo

numerical data presented in a structured format, typically related to performance metrics or statistics

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 8-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.8.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.8.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

 involuntary

-0.64

 exploited

-0.63

ande

-0.62

 withd

-0.62

 bondage

-0.61

 bonded

-0.60

 predatory

-0.59

 traders

-0.59

 privileged

-0.59

 governance

-0.58

POSITIVE LOGITS

icio

1.21

wikipedia

1.04

ItemTracker

0.94

assetsadobe

0.87

 Retrieved

0.85

Downloadha

0.82

MX

0.82

req

0.81

jpg

0.79

Ct

0.78

Activations Density 0.126%

percentages or numerical values presented in a specific format

numerical data presented in a structured format, typically related to performance metrics or statistics

No Comments

No Known Activations