INDEX

Explanations

specific sequences and combinations of letters, potentially related to programming or technical terms

oai_token-act-pair · gpt-3.5-turbo Triggered by @bot

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

PlotsExplanationShow Test FieldDefault Test Text

IFrame

Link

Not in Any Lists

No Comments

Head Attr Weights

0:0.08

1:0.03

2:0.26

3:0.07

4:0.15

5:0.06

6:0.03

7:0.02

8:0.09

9:0.08

10:0.06

11:0.02

Negative Logits

Applic

-1.24

querque

-1.21

actionDate

-1.19

zona

-1.14

 Contributions

-1.14

 Proced

-1.13

ikuman

-1.12

Lic

-1.10

 berth

-1.09

uckland

-1.09

POSITIVE LOGITS

sonian

1.36

 ourselves

1.22

nesday

1.14

ipeg

1.12

qt

1.11

anchester

1.10

SOURCE

1.07

schild

1.06

の�

1.05

inki

1.04

Activations Density 0.004%

specific sequences and combinations of letters, potentially related to programming or technical terms

No Comments

No Known Activations