nikmcfly/MiroFish-Offline

@kmckinley I try to do the same. Can you share more information how you pointed to LM Studio?

I wasn't able to get 3.5 to work within ollama, so instead I started qwen3.5-35b-a3b in LM Studio and pointed the project to use that. It worked well, though I did have to make a few changes to get it to work.

Increase the max_tokens. qwen3.5-35b-a3b is a thinking model so we needed to give it room for that.
```
def chat_json(
self,
messages: List[Dict[str, str]],
temperature: float = 0.3,
- max_tokens: int = 4096
+ max_tokens: int = 8192
) -> Dict[str, Any]:
```

Remove response_format: LM Studio doesn't support response_format: {"type": "json_object"} — only "json_schema" or "text". The chat_json method passes json_object mode which Ollama accepts but LM Studio rejects.
```
response = self.chat(
messages=messages,
temperature=temperature,
max_tokens=max_tokens,
- response_format={"type": "json_object"}
)
```

> 请问，qwen3.5的模型支持吗？

i would say possible but i have not got the base to work

this also is happening to me, i just have some doubts, could it be related to the default ollama service having a really small max token value. Also why qwen2.5 instead of qwen3.5?
Maybe in the yaml file set the context for the compose to launch ollama.

BTW, instruction on frontpage still mentioned Neojs 5.15, and should be updated to 5.18 for the docker run instruction.

Updated with Neo4j to 5.18. Didn't help.



Will investigate more later.

Thanks for looking into this. It did per screenshot, however nothing in it.



Step 2
NAME ID SIZE MODIFIED
nomic-embed-text:latest 0a109f422b47 274 MB 22 minutes ago
glm-4.7-flash:latest d1a8a26252f1 19 GB 3 days ago
gpt-oss:20b 17052f91a42e 13 GB 2 months ago
qwen3:latest 500a1f067a9f 5.2 GB 5 months ago
qwen3:30b 19e422b02313 18 GB 5 months ago
qwen3:8b 500a1f067a9f 5.2 GB 5 months ago
llava:13b 0d0eb4d7f485 8.0 GB 5 months ago
llava:latest 8dd30f6b0cb1 4.7 GB 5 months ago
llava-llama3:latest 44c161b1f465 5.5 GB 5 months ago
llama3.2:latest a80c4f17acd5 2.0 GB 5 months ago
llama3.2-vision:latest 6f2f9757ae97 7.8 GB 5 month...

Thanks for the report. The core issue is that your graph appears empty (0 nodes, 0 edges) — the simulation config generator can't proceed without graph data.

A few things to check:
1. Did Step 1 (graph building) complete successfully? Check the graph panel — do you see nodes and edges?
2. Is your Ollama embedding model loaded? Run `ollama list` and confirm you have an embedding model (e.g., `nomic-embed-text`)
3. The `created_at` property warning suggests the Neo4j schema may need updating — try running with Neo4j 5.18 (see PR #10)

Can you share:
- Your `.env` config (redact any API keys)
- Output of `ollama list`
- Whether Step 1 showed a populated graph

This will help narrow down whether it's a graph build failure or a config generation bug.