INDEX

Explanations

words related to criminal activities, particularly related to attempted crimes

oai_token-act-pair · gpt-3.5-turbo

repeated mentions of the word "attempted."

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 3-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.3.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.3.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

hander

-0.80

spr

-0.77

tal

-0.73

minus

-0.73

today

-0.73

houses

-0.70

front

-0.69

scene

-0.69

effect

-0.68

cum

-0.67

POSITIVE LOGITS

 unsuccessfully

0.99

Attempts

0.97

 attempts

0.93

 attempt

0.89

 attempted

0.85

heric

0.83

 carbohyd

0.82

ossibility

0.80

llor

0.77

 resusc

0.76

Activations Density 0.019%

words related to criminal activities, particularly related to attempted crimes

repeated mentions of the word "attempted."

No Comments

No Known Activations