INDEX

Explanations

No Explanations Found

New Auto-Interp

Top Features by Cosine Similarity

Configuration

ckkissane/attn-saes-gpt2-small-all-layers/gpt2-small_L7_Hcat_z_lr1.20e-03_l11.10e+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.7.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

0:0.11

1:0.05

2:0.05

3:0.09

4:0.08

5:0.09

6:0.07

7:0.04

8:0.21

9:0.06

10:0.04

11:0.04

Negative Logits

iability

-1.89

LR

-1.63

Slot

-1.61

fts

-1.59

ledge

-1.57

 secondly

-1.57

Nib

-1.55

icio

-1.51

aspers

-1.50

 Secondly

-1.49

POSITIVE LOGITS

 reconstructed

1.65

ographically

1.63

 veiled

1.60

owan

1.54

 scen

1.51

iann

1.50

 mayoral

1.50

stairs

1.47

enburg

1.46

 schizophren

1.46

Activations Density 0.000%

No Known Activations

This feature has no known activations.