openai/gpt-oss-20b · Discussions

#114 opened 4 months ago by

dsprag

Is it mandatory to install flash-attention for GPT-OSS?

#112 opened 4 months ago by

xiaotianyu2025

Unable to Structured output

9

#111 opened 4 months ago by

Arnab07

Running gpt-oss-20b on an RTX 4070 Ti (12GB) using Transformers

#110 opened 4 months ago by

Biiigstone

With openai harmony response, I cannot generate response.

#108 opened 4 months ago by

Noah0627

safe_serialization

#105 opened 4 months ago by

cuongdk253

NVIDIA L40S GPU's for MXFP4 quantization

#100 opened 4 months ago by

lordim

CUDA out of memory issues when running gptoss model on colab T4

#99 opened 4 months ago by

sumeetm

model quality issues

5

#92 opened 4 months ago by

TheBigBlockPC

The Abliterated version is the version we deserved.

😔 1

#91 opened 4 months ago by

Koitenshin

The censorship here is just... amazing.

👀 👍 8

#90 opened 4 months ago by

smokeofc

Does transformers utilize PyTorch SDPA's flash_attention for openai/gpt-oss-20b?

#89 opened 4 months ago by

NooBaymax

Quick Results: Testing Multilingual Capabilities via Code Generation from Ukrainian Competitive Programming Tasks

🚀 4

#87 opened 4 months ago by

anon-researcher-ua

How to turn off thinking mode

👍 🔥 7

8

#86 opened 4 months ago by

Gierry

I haven't seen a more useful model than this one.

#85 opened 4 months ago by

Maria99934

How to run it on macOS

❤️ 1

7

#84 opened 4 months ago by

kopyl

There is no spark.. only darkness

#83 opened 4 months ago by

JFE

tool calling not working as expected?

👍 2

10

#80 opened 4 months ago by

Spider-Jerusalem

SVG problem

#79 opened 4 months ago by

Alkohole

Honest Review of the Model

😎 👍 52

#78 opened 4 months ago by

MarinaraSpaghetti

behavior between GptOssExperts and Mxfp4GptOssExperts

#77 opened 4 months ago by

DaleMeng

'accelerate' missing in the model card.

👍 1

#76 opened 4 months ago by

Ina-Developer

Qwen3 beat gpt-oss with just 0.6B with good quality enough to be usable

👀 2

#75 opened 4 months ago by

yousef1727

triton_kernels and multiprocessing

#74 opened 4 months ago by

piotrm0

Thanks for the great model

😎 👀 4

#73 opened 4 months ago by

deniiiiiij

Running gpt-oss Without FlashAttention 3 – Any Alternatives to Ollama?

#72 opened 4 months ago by

shinho0902

getting curernt Date template issue with OLLAMA_ORIGINS="chrome-extension://*" OLLAMA_DEBUG=1 ollama serve

#71 opened 4 months ago by

lokesher

how to ignore the quantization

❤️ 👍 4

#70 opened 4 months ago by

zzbysd

Is there no usable PyTorch for Python 3.12?

#69 opened 4 months ago by

xiaotianyu2025

This is by far the worst model out there... Won't be using this. 🚮🗑️

🤯 ➕ 7

#67 opened 4 months ago by

UniversalLove333

question: setting reasoning effort

5

#66 opened 4 months ago by

TheBigBlockPC

🚀 Best Practices for Evaluating GPT-OSS Models: Speed & Benchmark Testing Guide

🤗 🔥 10

#64 opened 4 months ago by

Yunxz

information how to get it working on a 3090

👍 4

11

#63 opened 4 months ago by

TheBigBlockPC

It seems to be censored a bit too much.

🤯 👍 10

7

#62 opened 4 months ago by

petwoe

MXFP4 only runs on h100 or b100 or later versions,

😔 2

12

#61 opened 4 months ago by

kishan51

AIME 25 Accuracy Discrepancy for GPT-OSS-20B (Reasoning Effort=High)

#58 opened 4 months ago by

jiayi37u

🔥🔥🔥中文测评视频

#56 opened 4 months ago by

leo009

Report

🤝 1

#55 opened 4 months ago by

Mks1837

Thinking but no solution?

5

#54 opened 4 months ago by

ZeroWw

transformers + openai_gpt_oss on modal to run 👌🏻

👍 2

#51 opened 4 months ago by

weege007

self_attn.k_proj.bias are all 0 for all layers

#50 opened 4 months ago by

DaleMeng

no able to deploy using vLLM on A10G

#49 opened 4 months ago by

abhisskk

so much censorship

👍 24

9

#48 opened 4 months ago by

Utochi

How to set reasoning effort in the shown example?

👍 3

12

#47 opened 4 months ago by

TianheWu

When asked for a story with a gay dog character having a crush on a cat, the model only tells stories with straight characters, and cannot seem to correct itself.

😔 4

#46 opened 4 months ago by

CyroPaws

It's incredibly easy to get the model to spread false information defending conversion therapy for transgender people.

❤️ 👀 5

4

#45 opened 4 months ago by

CyroPaws

GGUFs of This model, tested, and conforms to tensor sizes - IQ4NL, Q5_1, MXFP4_MOE.gguf

#44 opened 4 months ago by

DavidAU

🚀[Fine-tuning] LoRA fine-tuning openai/gpt-oss-20b 👋

👀 👍 7