Bug: Crash attempting to run tiiuae/Falcon3-1B-Instruct-1.58bit-GGUF

[evangineer]

What happened?

So I'm running this on a Chromebook that llamafile reports as having the following CPU:
Intel Core i5-8200Y CPU @ 1.30GHz (skylake)

I have successfully run SmolLM2 (135M & 360M), TinyLlama 1.1B and Qwen 2.5 0.5B on this machine.

As is my wont, I used the following command line (after renaming the downloaded model to something less generic):

./llamafile --chat -f watchmen_chronology.txt -m Falcon3-1B-Instruct-1.58bit-GGUF

It crashed immediately with the following error:

llama.cpp/ggml.c:19663: GGML_ASSERT(0 <= info->type && info->type < GGML_TYPE_COUNT) failed

error: Uncaught SIGABRT (SI_TKILL) at 0x3e8000040ce on penguin pid 16590 tid 16590
  ./llamafile
  No error information
  Linux Cosmopolitan 4.0.2 MODE=x86_64; #1 SMP PREEMPT_DYNAMIC Thu, 7 Nov 2024 16:44:28 +0000 penguin 6.6.54-05528-gdd4efe62d86b

Version

llamafile v0.9.0

What operating system are you seeing the problem on?

Linux

Relevant log output

[evangineer]

I followed up by trying a less exotic Falcon 3 quant from tiiuae/Falcon3-1B-Instruct-GGUF.

Running ./llamafile --verbose --chat -f watchmen_chronology.txt -m Falcon3-1B-Instruct-q4_k_m.gguf crashed with the following error:

llama_model_load: error loading model: error loading model vocabulary: unknown pre-tokenizer type: 'falcon3'