INDEX

Explanations

terms related to specific "zones" or regions

oai_token-act-pair · gpt-3.5-turbo

references to various types of zones

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

NECT

-0.77

thora

-0.76

aldi

-0.76

anooga

-0.75

cho

-0.74

nant

-0.74

issance

-0.72

edes

-0.72

utive

-0.72

sonian

-0.72

POSITIVE LOGITS

 zone

0.95

 zones

0.93

 dwell

0.82

 Zone

0.75

 ozone

0.69

 closures

0.69

 boundaries

0.68

 alignment

0.68

 depletion

0.68

 boundary

0.67

Activations Density 0.026%

terms related to specific "zones" or regions

references to various types of zones

No Comments

No Known Activations