INDEX

Explanations

words related to negative outcomes or repercussions

oai_token-act-pair · gpt-3.5-turbo

references to the concept of consequences

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

PlotsExplanationShow Test FieldDefault Test Text

IFrame

Link

Not in Any Lists

No Comments

Negative Logits

 owned

-0.69

abee

-0.69

Bay

-0.67

 Sport

-0.65

qt

-0.65

berry

-0.64

Base

-0.64

wer

-0.63

Mus

-0.62

 Spot

-0.61

POSITIVE LOGITS

 consequences

3.67

 repercussions

2.96

 ramifications

2.69

 implications

2.36

 consequence

2.29

 reperc

1.98

 effects

1.65

effects

1.60

 dangers

1.56

 outcomes

1.53

Activations Density 0.020%

words related to negative outcomes or repercussions

references to the concept of consequences

No Comments

No Known Activations