INDEX

Explanations

phrases related to news outlets or journalistic articles

oai_token-act-pair · gpt-3.5-turbo

mentions of specific projects and agencies, particularly those starting with "Sentinel."

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 1-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.1.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.1.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

perm

-0.81

mph

-0.76

akery

-0.75

places

-0.75

sm

-0.74

shop

-0.73

shared

-0.73

roma

-0.73

sen

-0.72

asia

-0.71

POSITIVE LOGITS

 Sentinel

1.57

inel

0.92

 Warden

0.81

 Extensions

0.80

 Advocate

0.78

 Superior

0.76

 Pulse

0.76

 Feedback

0.74

 CrossRef

0.74

 Tracker

0.73

Activations Density 0.011%

phrases related to news outlets or journalistic articles

mentions of specific projects and agencies, particularly those starting with "Sentinel."

No Comments

No Known Activations