INDEX

Explanations

names or fragments of names, especially those with a strong emphasis on the letter T

oai_token-act-pair · gpt-3.5-turbo

proper nouns and names

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 5-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.5.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.5.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

 STATS

-0.70

 innocence

-0.63

 sanct

-0.63

 succession

-0.61

 preservation

-0.61

 VIDE

-0.61

 sanity

-0.60

 envelope

-0.60

 attendant

-0.59

 meditation

-0.58

POSITIVE LOGITS

oshenko

1.11

edo

1.00

hiba

0.87

asury

0.86

nel

0.85

oval

0.85

arkin

0.85

emouth

0.84

ulhu

0.84

obl

0.82

Activations Density 0.126%

names or fragments of names, especially those with a strong emphasis on the letter T

proper nouns and names

No Comments

No Known Activations