INDEX

Explanations

instances where something is being taken to a more intense level

oai_token-act-pair · gpt-3.5-turbo

the recurrence of the term "further" in contexts suggesting continuation or escalation

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

odor

-0.78

liest

-0.78

ldom

-0.69

amine

-0.64

ipe

-0.63

Ely

-0.61

ertodd

-0.60

 catcher

-0.60

cko

-0.60

essee

-0.59

POSITIVE LOGITS

ance

0.95

 refine

0.82

 than

0.80

 largeDownload

0.75

 refinement

0.74

 distances

0.74

 refining

0.73

foundland

0.72

 refined

0.71

 subdiv

0.71

Activations Density 0.029%

instances where something is being taken to a more intense level

the recurrence of the term "further" in contexts suggesting continuation or escalation

No Comments

No Known Activations