INDEX

Explanations

phrases related to reported incidents or criminal activities

oai_token-act-pair · gpt-3.5-turbo

references to a specific event or occurrence, labeled as an "incident."

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 0-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.0.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.0.hook_resid_pre

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

YES

-0.78

ellen

-0.75

States

-0.64

marks

-0.62

 Strongh

-0.61

esters

-0.61

 Lists

-0.60

 slots

-0.60

 coffers

-0.60

states

-0.60

POSITIVE LOGITS

 incident

3.42

 Incident

2.41

 incidents

2.14

 altercation

1.78

 accident

1.74

 occurrence

1.37

 event

1.35

 encounter

1.35

 anecdote

1.33

 ordeal

1.30

Activations Density 0.016%

phrases related to reported incidents or criminal activities

references to a specific event or occurrence, labeled as an "incident."

No Comments

No Known Activations

phrases related to reported incidents or criminal activities

references to a specific event or occurrence, labeled as an "incident."

No Comments

No Known Activations