INDEX

Explanations

mentions of security, intelligence, and investigations

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

names and titles of individuals, particularly those involved in government or intelligence

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 12-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.11.hook_resid_post

Prompts (Dashboard)

24,576 prompts, 128 tokens each

Dataset (Dashboard)

Skylion007/openwebtext

Features

24,576

Data Type

torch.float32

Hook Point

blocks.11.hook_resid_post

Architecture

standard

Context Size

128

Dataset

Skylion007/openwebtext

Hook Point Layer

Activation Function

relu

Embeds

IFrame

Link

Not in Any Lists

No Comments

Negative Logits

 Samar

-0.70

MAP

-0.68

 bride

-0.68

ortment

-0.68

 Colomb

-0.68

Quantity

-0.65

Buyable

-0.65

 curing

-0.65

 amput

-0.65

akeru

-0.65

POSITIVE LOGITS

 Nunes

1.48

 subpoena

1.45

 subpoen

1.42

 Oversight

1.37

 Ranking

1.34

 Subcommittee

1.33

ittee

1.25

 subcommittee

1.25

 committee

1.24

 Schiff

1.23

Activations Density 0.335%

mentions of security, intelligence, and investigations

names and titles of individuals, particularly those involved in government or intelligence

No Comments

No Known Activations