INDEX

Explanations

distinguishing terms related to specific events or settings called "Round"

oai_token-act-pair · gpt-3.5-turbo

references to rounds or series of events, likely in a competitive or structured context

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 0-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.0.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.0.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

İĭ

-0.82

itability

-0.65

 disav

-0.64

lov

-0.64

 experien

-0.63

cop

-0.63

artifacts

-0.62

 dele

-0.62

 attribut

-0.62

 sensit

-0.61

POSITIVE LOGITS

 Round

3.90

Round

3.21

 Rounds

2.35

 round

2.17

round

1.88

 rounds

1.72

 Quarter

1.27

 Roundup

1.10

rounder

1.09

 Semi

1.08

Activations Density 0.010%

distinguishing terms related to specific events or settings called "Round"

references to rounds or series of events, likely in a competitive or structured context

No Comments

No Known Activations