INDEX

Explanations

No Explanations Found

New Auto-Interp

Top Features by Cosine Similarity

Configuration

ckkissane/attn-saes-gpt2-small-all-layers/gpt2-small_L5_Hcat_z_lr1.20e-03_l11.00e+00_ds49152_bs4096_dc1.00e-06_rsanthropic_rie25000_nr4_v9.pt

Prompts (Dashboard)

36,864 prompts, 128 tokens each

Dataset (Dashboard)

Skylion007/openwebtext

Features

49,152

Data Type

float32

Hook Name

blocks.5.attn.hook_z

Hook Layer

Architecture

standard

Context Size

128

Dataset

Skylion007/openwebtext

Activation Function

relu

Embeds

IFrame

Link

Head Attr Weights

0:0.08

1:0.04

2:0.07

3:0.07

4:0.09

5:0.09

6:0.07

7:0.08

8:0.08

9:0.09

10:0.09

11:0.08

Negative Logits

 conduc

-2.10

 combatants

-1.69

 manpower

-1.68

 instit

-1.67

 administr

-1.61

 prosec

-1.60

 incarcer

-1.56

 coerc

-1.54

 youths

-1.48

 solicit

-1.48

POSITIVE LOGITS

ocious

1.80

NAS

1.73

pent

1.68

nect

1.63

魔

1.62

ARB

1.60

Letter

1.59

feature

1.58

found

1.56

float

1.54

Activations Density 0.000%

No Known Activations

This feature has no known activations.