# \#llm

**URL:** https://discuss.flower.ai/tag/llm/32.md

[Latest](https://discuss.flower.ai/latest.md) · [Categories](https://discuss.flower.ai/categories.md) · [Tags](https://discuss.flower.ai/tags.md)

---

## [Fine-tuning LLMs with Flower](https://discuss.flower.ai/t/fine-tuning-llms-with-flower/1031)

<div class="topic-metadata">

**Author:** [@damjimenezgu](https://discuss.flower.ai/u/damjimenezgu)\
**Replies:** 2\
**Last updated:** [September 11, 2025, 1:10pm UTC](https://discuss.flower.ai/t/fine-tuning-llms-with-flower/1031 "2025-09-11T13:10:16Z")

</div>

Hi! I’m a newbie working on fine-tuning LLMs for a chatbot in Italian. I have followed this guide. The training goes (apparently) well, but when I check the answers of the model, they are not similar to the training set.…

---

## ["RBAC: access denied" error running FlowerTune LLM example in deployment mode on multiple machines](https://discuss.flower.ai/t/rbac-access-denied-error-running-flowertune-llm-example-in-deployment-mode-on-multiple-machines/1011)

<div class="topic-metadata">

**Author:** [@griffith](https://discuss.flower.ai/u/griffith)\
**Replies:** 1\
**Last updated:** [June 17, 2025, 8:47am UTC](https://discuss.flower.ai/t/rbac-access-denied-error-running-flowertune-llm-example-in-deployment-mode-on-multiple-machines/1011 "2025-06-17T08:47:28Z")

</div>

I’m working with the FlowerTune LMM example in a KubeFlow environment. I can get it to run in simulation mode and I can get it to run in deployment mode on the same node/pod, but I get the error below when I try to start…

---

## [Methods for Improving Inference Speed During LLM Fine-Tuning](https://discuss.flower.ai/t/methods-for-improving-inference-speed-during-llm-fine-tuning/376)

<div class="topic-metadata">

**Author:** [@jincheol](https://discuss.flower.ai/u/jincheol)\
**Replies:** 0\
**Last updated:** [October 24, 2024, 2:40pm UTC](https://discuss.flower.ai/t/methods-for-improving-inference-speed-during-llm-fine-tuning/376 "2024-10-24T14:40:11Z")

</div>

Could you suggest any ways to improve the response speed when using the results of fine-tuning an LLM model with Flower for inference tasks? Specifically, are there any techniques that can be applied during the training …

---

## [Announcing FlowerTune LLM Leaderboard](https://discuss.flower.ai/t/announcing-flowertune-llm-leaderboard/359)

<div class="topic-metadata">

**Author:** [@julian](https://discuss.flower.ai/u/julian)\
**Replies:** 0\
**Last updated:** [October 22, 2024, 3:56pm UTC](https://discuss.flower.ai/t/announcing-flowertune-llm-leaderboard/359 "2024-10-22T15:56:49Z")

</div>

We are excited to launch the FlowerTune LLM Leaderboard ! :rocket: In this initiative, we provide a complete pipeline for the federated fine-tuning a pre-trained Mistral-7B across 4 tasks with model performance measu…