# Is it possible to run tiny llama using relay?

**URL:** <https://discuss.tvm.apache.org/t/is-it-possible-to-run-tiny-llama-using-relay/18070>\
**Category:** Questions\
**Created:** [February 12, 2025, 6:44pm UTC](https://discuss.tvm.apache.org/t/is-it-possible-to-run-tiny-llama-using-relay/18070 "2025-02-12T18:44:08Z")\
**Posts on this page:** 2\
**Page:** 1

<div class="post-metadata">

**Author:** ![Irfan](https://discuss.tvm.apache.org/user_avatar/discuss.tvm.apache.org/irfan/32/5999_2.png) [@Irfan](https://discuss.tvm.apache.org/u/Irfan)\
**Post date:** [February 12, 2025, 6:44pm UTC](https://discuss.tvm.apache.org/t/is-it-possible-to-run-tiny-llama-using-relay/18070/1 "2025-02-12T18:44:08Z")

</div>

If some how we make the llama model static instead of dynamic is it possible to run it using the graph executor after converting to relay module? I have been through a documentation of relax it is possible there but I need to run it with relay instead. Any suggestion or guideline would be greatly appreciated. Thank You!

---

<div class="post-metadata">

**Author:** ![Hzfengsy](https://discuss.tvm.apache.org/user_avatar/discuss.tvm.apache.org/hzfengsy/32/2961_2.png) [@Hzfengsy](https://discuss.tvm.apache.org/u/Hzfengsy)\
**Post date:** [February 13, 2025, 10:46am UTC](https://discuss.tvm.apache.org/t/is-it-possible-to-run-tiny-llama-using-relay/18070/2 "2025-02-13T10:46:37Z")

</div>

Relay does not support dynamic-shape and kv-cache. So it’s hard to run through Relay.

We are phasing out Relay, please try moving towards relax 🙂
