INDEX

Explanations

phrases indicating change or significant events

oai_token-act-pair · gpt-3.5-turbo

consequences or events that have transpired

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

 notoriously

-0.78

 intended

-0.75

ommel

-0.72

 preferred

-0.72

 tack

-0.72

 ornament

-0.71

 compar

-0.70

 finer

-0.70

 concess

-0.70

unte

-0.69

POSITIVE LOGITS

 Suddenly

1.35

 Slowly

1.24

 Soon

1.14

 Thousands

1.10

 Eventually

1.09

 Within

1.08

 Immediately

1.06

 Gone

1.05

 Hundreds

1.04

 Unable

1.03

Activations Density 0.530%

phrases indicating change or significant events

consequences or events that have transpired

No Comments

No Known Activations