INDEX

Explanations

terms related to explosive events or actions

oai_token-act-pair · gpt-3.5-turbo

occurrences of the word "blast" 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 @ 0-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.0.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.0.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

vier

-0.89

Ã©t

-0.78

 discont

-0.72

aniel

-0.66

ais

-0.66

iture

-0.66

Relations

-0.65

 Kenn

-0.64

relations

-0.64

ns

-0.64

POSITIVE LOGITS

 blast

3.98

 blasts

2.56

blast

1.93

 Blast

1.84

 blasting

1.75

 explosion

1.61

 explosions

1.34

 blasted

1.33

 slam

1.32

 blaze

1.27

Activations Density 0.013%

terms related to explosive events or actions

occurrences of the word "blast" in various contexts

No Comments

No Known Activations