INDEX

Explanations

phrases related to legal proceedings and statements

New Auto-Interp

Top Features by Cosine Similarity

Configuration

ckkissane/attn-saes-gpt2-small-all-layers/gpt2-small_L9_Hcat_z_lr1.20e-03_l11.20e+00_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.9.attn.hook_z

Hook Layer

Architecture

standard

Context Size

128

Dataset

Skylion007/openwebtext

Activation Function

relu

Embeds

IFrame

Link

Head Attr Weights

0:0.05

1:0.02

2:0.05

3:0.20

4:0.03

5:0.06

6:0.04

7:0.05

8:0.02

9:0.02

10:0.39

11:0.02

Negative Logits

 temptation

-2.43

])

-2.16

 REALLY

-2.09

 experimented

-1.99

 dared

-1.94

 incess

-1.93

 tempted

-1.83

 taboo

-1.82

 suspicion

-1.81

 bothers

-1.77

POSITIVE LOGITS

Comment

2.50

endum

2.43

 update

2.38

reply

2.37

 roundup

2.31

 statement

2.20

pedia

2.16

 Update

2.14

 Transcript

2.13

respond

2.13

Activations Density 0.101%

phrases related to legal proceedings and statements

No Known Activations