INDEX

Explanations

keywords related to official titles or positions, particularly those related to governmental or organizational roles

oai_token-act-pair · gpt-3.5-turbo

occurrences of the term "Com" in various contexts

oai_token-act-pair · gpt-4o-mini Triggered by @bot

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 8-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.8.hook_resid_pre

Prompts (Dashboard)

24,576 prompts, 128 tokens each

Dataset (Dashboard)

Skylion007/openwebtext

Features

24,576

Data Type

torch.float32

Hook Point

blocks.8.hook_resid_pre

Architecture

standard

Context Size

128

Dataset

Skylion007/openwebtext

Hook Point Layer

Activation Function

relu

Embeds

PlotsExplanationShow Test FieldDefault Test Text

IFrame

Link

Not in Any Lists

No Comments

Negative Logits

 Kubrick

-0.75

 harbor

-0.68

 Icelandic

-0.65

 Refuge

-0.65

 infringing

-0.64

 halfway

-0.64

 repetitive

-0.62

 Mara

-0.60

 Fukushima

-0.60

 Ducks

-0.59

POSITIVE LOGITS

pletion

1.26

ptroller

1.24

mented

1.23

fortable

1.23

pleting

1.22

plex

1.17

plement

1.14

puters

1.13

pos

1.08

plet

1.07

Activations Density 0.022%

keywords related to official titles or positions, particularly those related to governmental or organizational roles

occurrences of the term "Com" in various contexts

No Comments

No Known Activations