INDEX

Explanations

words related to intense emotional reactions or physical situations

oai_token-act-pair · gpt-3.5-turbo

instances of the word "heated" in various contexts related to intensity or temperature

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 1-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.1.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.1.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

ateral

-0.77

roma

-0.73

ogo

-0.71

ment

-0.70

 Commando

-0.69

ystem

-0.66

ops

-0.65

 Grants

-0.65

padding

-0.65

lus

-0.65

POSITIVE LOGITS

 heated

1.35

 heating

0.94

 overhe

0.91

 heats

0.91

ãĤ¦ãĤ¹

0.84

 cooled

0.83

 hotter

0.83

 uproar

0.79

 raged

0.77

 temperatures

0.75

Activations Density 0.007%

words related to intense emotional reactions or physical situations

instances of the word "heated" in various contexts related to intensity or temperature

No Comments

No Known Activations