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

Not in Any Lists

No Comments

Head Attr Weights

0:0.08

1:0.08

2:0.09

3:0.07

4:0.08

5:0.07

6:0.08

7:0.06

8:0.09

9:0.08

10:0.09

11:0.09

Negative Logits

 interfered

-1.80

 presided

-1.78

 coached

-1.60

unia

-1.59

……

-1.57

王

-1.51

 unilaterally

-1.50

 raped

-1.49

rape

-1.48

illegal

-1.48

POSITIVE LOGITS

jam

1.82

 Photographer

1.68

 PLoS

1.58

 Stam

1.54

icent

1.51

 enthusi

1.50

1.49

racuse

1.49

 herald

1.48

NCT

1.46

Activations Density 0.000%

No Known Activations

This feature has no known activations.