INDEX

Explanations

sentences with a strong emphasis on prohibition or warning against certain actions

oai_token-act-pair · gpt-3.5-turbo

phrases indicating conditions or restrictions related to actions or behaviors

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

 Voting

-0.68

 Borough

-0.64

 Room

-0.61

 boarding

-0.60

Eye

-0.60

 Meadow

-0.60

hips

-0.59

 nerv

-0.59

 Cards

-0.59

 Upper

-0.58

POSITIVE LOGITS

 violates

1.09

 occurs

1.05

 involves

1.04

 contradicts

0.99

 satisfies

0.97

 accompanies

0.95

 exceeds

0.94

 resembles

0.94

 contains

0.93

 entails

0.93

Activations Density 0.179%

sentences with a strong emphasis on prohibition or warning against certain actions

phrases indicating conditions or restrictions related to actions or behaviors

No Comments

No Known Activations