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

1:0.10

2:0.08

3:0.06

4:0.07

5:0.08

6:0.08

7:0.09

8:0.07

9:0.07

10:0.08

11:0.09

Negative Logits

shoot

-1.97

compatible

-1.93

lishes

-1.90

correct

-1.83

BST

-1.82

 recess

-1.80

 discredited

-1.76

business

-1.70

 copying

-1.68

iliated

-1.67

POSITIVE LOGITS

ofi

2.06

Redditor

1.98

enery

1.88

 Rasm

1.88

Developer

1.88

umar

1.86

�

1.84

Uh

1.81

 enthusi

1.79

 Authors

1.78

Activations Density 0.000%

No Known Activations

This feature has no known activations.