INDEX

Explanations

medical terms and scientific concepts

oai_token-act-pair · gpt-3.5-turbo Triggered by @bot

specific phrases and punctuation that indicate lists or enumerations

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 2-att-kk

Configuration

ckkissane/attn-saes-gpt2-small-all-layers/gpt2-small_L2_Hcat_z_lr1.20e-03_l11.00e+00_ds24576_bs4096_dc1.00e-06_rsanthropic_rie25000_nr4_v4.pt

Prompts (Dashboard)

36,864 prompts, 128 tokens each

Dataset (Dashboard)

Skylion007/openwebtext

Features

24,576

Data Type

float32

Hook Name

blocks.2.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

0:0.04

1:0.02

2:0.13

3:0.06

4:0.03

5:0.13

6:0.04

7:0.08

8:0.04

9:0.14

10:0.16

11:0.07

Negative Logits

�

-1.16

IDENT

-1.12

enium

-1.09

agon

-1.09

erest

-1.03

ISION

-0.99

ennial

-0.96

coat

-0.95

iHUD

-0.95

 thug

-0.94

POSITIVE LOGITS

xtap

1.23

conservancy

1.22

STEM

1.08

 respectively

1.08

 RPGs

1.02

 alike

0.99

 lined

0.98

 totaling

0.97

etting

0.95

 boycot

0.94

Activations Density 0.251%

medical terms and scientific concepts

specific phrases and punctuation that indicate lists or enumerations

No Comments

No Known Activations