INDEX

Explanations

phrases that encourage recalling past events or information

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

New Auto-Interp

Top Features by Cosine Similarity

Configuration

ckkissane/attn-saes-gpt2-small-all-layers/gpt2-small_L3_Hcat_z_lr1.20e-03_l19.00e-01_ds24576_bs4096_dc1.00e-06_rsanthropic_rie25000_nr4_v9.pt

Prompts (Dashboard)

36,864 prompts, 128 tokens each

Dataset (Dashboard)

Skylion007/openwebtext

Features

24,576

Data Type

float32

Hook Name

blocks.3.attn.hook_z

Hook Layer

Architecture

standard

Context Size

128

Dataset

Skylion007/openwebtext

Activation Function

relu

Embeds

IFrame

Link

Not in Any Lists

No Comments

Head Attr Weights

0:0.01

1:0.06

2:0.14

3:0.05

4:0.01

5:0.02

6:0.14

7:0.08

8:0.07

9:0.16

10:0.07

11:0.13

Negative Logits

thro

-1.06

ansson

-1.02

UGC

-1.01

EStreamFrame

-1.00

iew

-1.00

urate

-1.00

Europe

-1.00

incial

-0.98

Í

-0.97

BuyableInstoreAndOnline

-0.97

POSITIVE LOGITS

 yourselves

1.11

atorium

1.11

 deleted

1.09

 Remastered

1.07

 tapes

1.00

 covenant

0.98

 Incident

0.93

 typo

0.93

 Watergate

0.92

Activations Density 0.013%

phrases that encourage recalling past events or information

No Comments

No Known Activations