runtime error

Exit code: 1. Reason: g-Q4_K_M.gguf: 0%| | 0.00/731M [00:00<?, ?B/s] ./LFM2.5-1.2B-Thinking-Q4_K_M.gguf: 65%|██████▌ | 476M/731M [00:01<00:00, 351MB/s] ./LFM2.5-1.2B-Thinking-Q4_K_M.gguf: 100%|██████████| 731M/731M [00:02<00:00, 356MB/s] llama_context: non-unified KV cache requires ggml_set_rows() - forcing unified KV cache llama_context: n_ctx_per_seq (2048) < n_ctx_train (128000) -- the full capacity of the model will not be utilized llama_kv_cache_unified: LLAMA_SET_ROWS=0, using old ggml_cpy() method for backwards compatibility [INFO] Model loaded. [INFO] Warming up... [INFO] Warmup done. Traceback (most recent call last): File "/app/app.py", line 134, in <module> for _ in llm.create_chat_completion( ~~~~~~~~~~~~~~~~~~~~~~~~~~^ messages=[{"role": "user", "content": "Count from 1 to 10."}], ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^ max_tokens=32, temperature=0.0, stream=True, ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^ ): ^ File "/usr/local/lib/python3.13/site-packages/llama_cpp/llama_chat_format.py", line 321, in _convert_text_completion_chunks_to_chat for i, chunk in enumerate(chunks): ~~~~~~~~~^^^^^^^^ File "/usr/local/lib/python3.13/site-packages/llama_cpp/llama.py", line 1322, in _create_completion for token in self.generate( ~~~~~~~~~~~~~^ prompt_tokens, ^^^^^^^^^^^^^^ ...<14 lines>... grammar=grammar, ^^^^^^^^^^^^^^^^ ): ^ File "/usr/local/lib/python3.13/site-packages/llama_cpp/llama.py", line 914, in generate self.eval(tokens) ~~~~~~~~~^^^^^^^^ File "/usr/local/lib/python3.13/site-packages/llama_cpp/llama.py", line 648, in eval self._ctx.decode(self._batch) ~~~~~~~~~~~~~~~~^^^^^^^^^^^^^ File "/usr/local/lib/python3.13/site-packages/llama_cpp/_internals.py", line 327, in decode raise RuntimeError(f"llama_decode returned {return_code}") RuntimeError: llama_decode returned -1

Container logs: