INDEX

Explanations

phrases related to identity and authority, particularly in the context of news reporting and personal accounts

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_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

Not in Any Lists

No Comments

Head Attr Weights

0:0.10

1:0.02

2:0.26

3:0.15

4:0.04

5:0.06

6:0.03

7:0.05

8:0.08

9:0.02

10:0.08

11:0.05

Negative Logits

 planners

-3.01

 architects

-2.74

 productivity

-2.74

 mobility

-2.67

 treadmill

-2.63

 optimal

-2.57

 glide

-2.57

 innovations

-2.47

 adaptive

-2.45

liv

-2.40

POSITIVE LOGITS

 apologised

3.61

TMZ

3.57

 allegations

3.57

 allegation

3.54

 allege

3.43

 suspicions

3.42

 alleges

3.40

 accusation

3.39

 accusing

3.36

 slander

3.35

Activations Density 0.713%

phrases related to identity and authority, particularly in the context of news reporting and personal accounts

No Comments

No Known Activations