97 5 197

Csaba Kecskemeti PRO

csabakecskemeti

https://devquasar.com/

csabakecskemeti

AI & ML interests

None yet

Recent Activity

updated a model about 2 hours ago

DevQuasar/mistralai.Mistral-Large-3-675B-Instruct-2512-GGUF

updated a model about 4 hours ago

DevQuasar/nvidia.Nemotron-Content-Safety-Reasoning-4B-GGUF

updated a model about 4 hours ago

DevQuasar/nvidia.Nemotron-Content-Safety-Reasoning-4B-GGUF

View all activity

Organizations

Posts 32

Post

908

FYI: Mistral.Ministral-3 dequantizer FP8->BF16

https://github.com/csabakecskemeti/ministral-3_dequantizer_fp8-bf16

(The instruct model weights are in FP8)

Post

1998

Looking for some help to test an INT8 Deepseek 3.2:
SGLang supports Channel wise INT8 quants on CPUs with AMX instructions (Xeon 5 and above AFAIK)
https://lmsys.org/blog/2025-07-14-intel-xeon-optimization/

Currently uploading an INT8 version of Deepseek 3.2 Speciale:
DevQuasar/deepseek-ai.DeepSeek-V3.2-Speciale-Channel-INT8

I cannot test this I'm on AMD
"AssertionError: W8A8Int8LinearMethod on CPU requires that CPU has AMX support"
(I assumed it can fall back to some non optimized kernel but seems not)

If anyone with the required resources (Intel Xeon 5/6 + ~768-1TB ram) can help to test this that would be awesome.

If you have hints how to make this work on AMD Threadripper 7000 Pro series please guide me.

Thanks all!

View all Posts

Collections 2

spaces 2

Mcp Course Gradio Client-smolagent

📊

Answer questions using MCP tools

Mcp Sentiment

⚡

HF MCP Course mcp-sentiment

models 58

datasets 0

None public yet

Csaba Kecskemeti PRO

AI & ML interests

Recent Activity

Organizations

Posts 32

Collections 2

XQuant: Breaking the Memory Wall for LLM Inference with KV Cache Rematerialization

city96/Wan2.1-T2V-14B-gguf

XQuant: Breaking the Memory Wall for LLM Inference with KV Cache Rematerialization

city96/Wan2.1-T2V-14B-gguf

spaces 2

Mcp Course Gradio Client-smolagent

Mcp Sentiment

models 58

csabakecskemeti/dq-gpt2-instruct-exp1

csabakecskemeti/PrimeIntellect.INTELLECT-3-GGUF

csabakecskemeti/deepseek-ai.DeepSeek-Math-V2-GGUF

csabakecskemeti/inclusionAI.Ling-1T-GGUF

csabakecskemeti/cerebras.Qwen3-Coder-REAP-25B-A3B-GGUF

csabakecskemeti/Salesforce.GTA1-32B-GGUF

csabakecskemeti/GAIR.LIMI-GGUF

csabakecskemeti/Yehor.kulyk-uk-en-GGUF

csabakecskemeti/Yehor.kulyk-en-uk-GGUF

csabakecskemeti/Locutusque.DareQwen-2.5-7B-GGUF

datasets 0

Csaba Kecskemeti PRO

AI & ML interests

Recent Activity

Organizations

Posts 32

Collections 2

spaces 2 Sort: Recently updated

Mcp Course Gradio Client-smolagent

Mcp Sentiment

models 58 Sort: Recently updated

datasets 0

spaces 2

models 58