INDEX

Explanations

phrases related to concealing or hiding specific information

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

New Auto-Interp

Top Features by Cosine Similarity

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

0:0.06

1:0.04

2:0.12

3:0.12

4:0.08

5:0.07

6:0.03

7:0.03

8:0.15

9:0.14

10:0.07

11:0.03

Negative Logits

ardless

-1.28

reau

-1.28

aughed

-1.26

IPM

-1.25

 surveyed

-1.23

untarily

-1.22

 booted

-1.22

untary

-1.21

 adjusted

-1.21

 resumed

-1.17

POSITIVE LOGITS

Versions

1.43

circ

1.42

 clues

1.39

 mysteries

1.36

Orig

1.35

ographical

1.35

genre

1.32

 workings

1.31

 distinctions

1.28

 interpretations

1.27

Activations Density 0.006%

phrases related to concealing or hiding specific information

No Comments

No Known Activations