INDEX

Explanations

references to television shows or news segments

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 8-res_fs24576-jb

Configuration

jbloom/GPT2-Small-Feature-Splitting-Experiment-Layer-8/blocks.8.hook_resid_pre_24576

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

 packing

-0.75

etsy

-0.68

 scrib

-0.67

ometric

-0.66

 bloom

-0.66

 shipping

-0.66

alties

-0.65

 extinct

-0.65

 curs

-0.64

idas

-0.64

POSITIVE LOGITS

 Tonight

1.16

 radio

1.12

 MSNBC

1.11

 interviewer

1.11

 broadcasts

1.08

 television

1.08

 aired

1.07

Radio

1.07

radio

1.05

 airs

1.05

Activations Density 0.236%

references to television shows or news segments

No Comments

No Known Activations