INDEX

Explanations

language indicating progression, advancement, or intensification

oai_token-act-pair · gpt-3.5-turbo

instances of the word "further" in various contexts

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 11-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.11.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.11.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

pei

-0.77

pas

-0.70

orno

-0.62

EA

-0.61

 Faces

-0.61

-0.60

pty

-0.60

ldom

-0.60

 Thousand

-0.59

Pry

-0.59

POSITIVE LOGITS

 complicate

0.94

 refine

0.91

 largeDownload

0.88

ance

0.82

rawdownloadcloneembedreportprint

0.81

 convol

0.80

 refinement

0.80

 deterior

0.80

Increases

0.78

 deterioration

0.78

Activations Density 0.031%

language indicating progression, advancement, or intensification

instances of the word "further" in various contexts

No Comments

No Known Activations