INDEX

Explanations

food-related terms along with references to specific objects or locations

New Auto-Interp

Top Features by Cosine Similarity

Configuration

ckkissane/attn-saes-gpt2-small-all-layers/gpt2-small_L2_Hcat_z_lr1.20e-03_l11.00e+00_ds24576_bs4096_dc1.00e-06_rsanthropic_rie25000_nr4_v4.pt

Prompts (Dashboard)

36,864 prompts, 128 tokens each

Dataset (Dashboard)

Skylion007/openwebtext

Features

24,576

Data Type

float32

Hook Name

blocks.2.attn.hook_z

Hook Layer

Architecture

standard

Context Size

128

Dataset

Skylion007/openwebtext

Activation Function

relu

Embeds

IFrame

Link

Head Attr Weights

0:0.07

1:0.04

2:0.11

3:0.08

4:0.03

5:0.16

6:0.03

7:0.05

8:0.10

9:0.11

10:0.12

11:0.04

Negative Logits

cipled

-1.59

ulative

-1.35

 reconciliation

-1.21

ploy

-1.18

 inconsistency

-1.15

 alliance

-1.12

 Analytics

-1.11

channelAvailability

-1.10

 absentee

-1.09

 sched

-1.09

POSITIVE LOGITS

ensis

1.41

ride

1.28

itaire

1.23

resses

1.18

icles

1.16

arettes

1.15

↵

1.14

oops

1.13

hers

1.13

ques

1.12

Activations Density 0.946%

food-related terms along with references to specific objects or locations

No Known Activations