INDEX

Explanations

phrases related to disappearance or removal

oai_token-act-pair · gpt-3.5-turbo

terms related to removal or loss

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

weeney

-0.66

rouse

-0.65

 Rhythm

-0.64

venants

-0.62

 Seas

-0.61

 Investigative

-0.59

 Strikes

-0.58

Tang

-0.57

monary

-0.56

ebus

-0.55

POSITIVE LOGITS

 altogether

1.38

 entirely

1.08

 obsolete

0.98

 abruptly

0.90

 anytime

0.81

 undone

0.78

:(

0.78

 completely

0.77

 replaced

0.75

 outright

0.74

Activations Density 0.213%

phrases related to disappearance or removal

terms related to removal or loss

No Comments

No Known Activations