gguf_init_from_file: invalid magic characters - Fine Tuned Model -

[dynamite9999]

Hello,
I followed the sample colab notebook and fine tuned - "unsloth/Meta-Llama-3.1-8B-bnb-4bit" model.

I used the latest llama.cpp compiled with flags cmake -B build -DGGML_CUDA=ON -DGGML_CUDA_ENABLE_UNIFIED_MEMORY=1

It generated the gguf file no problem, but when I tried to use the generated gguf I got this error:
c$ ./main -m ./models/unsloth.Q4_K_M.gguf -p "hello"
Log start
main: build = 3482 (e54c35e4)
main: built with cc (Ubuntu 11.4.0-1ubuntu1~22.04) 11.4.0 for x86_64-linux-gnu
main: seed = 1735960410
gguf_init_from_file: invalid magic characters ''
llama_model_load: error loading model: llama_model_loader: failed to load model from ./models/unsloth.Q4_K_M.gguf

llama_load_model_from_file: failed to load model
llama_init_from_gpt_params: error: failed to load model './models/unsloth.Q4_K_M.gguf'
main: error: unable to load model

Here is the first few bytes of the generated gguf file, any experts see any issues with the generated gguf ?

(netai) d@d:/hp/NetAnalytics/dev/netai/syslog/syslog_scraper_netai/t80/rc$ hexdump -C ./models/unsloth.Q4_K_M.gguf | head -n 10
00000000 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 |................|
*
00777e20 00 00 80 3f 00 00 80 3f 00 00 80 3f 00 00 80 3f |...?...?...?...?|
*
00777e50 00 00 80 3f 00 00 80 3f 00 00 80 3f c2 5e d3 3f |...?...?...?.^.?|
00777e60 6f b4 52 40 ee aa 1a 41 00 00 00 42 00 00 00 42 |[email protected]|
00777e70 00 00 00 42 00 00 00 42 00 00 00 42 00 00 00 42 |...B...B...B...B|
*
00777ea0 dc 5a 06 ac 97 b8 0f 2a 94 88 da 3f c1 7d 8e 71 |.Z.....*...?.}.q|
00777eb0 f4 a2 db 17 fe 31 75 eb 87 6f 00 0b 58 39 54 44 |.....1u..o..X9TD|
(netai) d@d:/hp/NetAnalytics/dev/netai/syslog/syslog_scraper_netai/t80/rc$

Any ideas on how to figure out how to start debugging ?

[dynamite9999]

More info.
The 16Bit generated model works fine. During the process of constructing the quanitzed 4bit unsloth/llama3.2 model, it creates a 16 Brain Float model, and it works. Here is some more info
model_q4_k_m$ ls -lrt
total 7002584
-rw-rw-r-- 1 d d 54628 Jan 3 19:10 tokenizer_config.json
-rw-rw-r-- 1 d d 454 Jan 3 19:10 special_tokens_map.json
-rw-rw-r-- 1 d d 17209920 Jan 3 19:10 tokenizer.json
-rw-rw-r-- 1 d d 994 Jan 3 19:10 config.json
-rw-rw-r-- 1 d d 234 Jan 3 19:10 generation_config.json
-rw-rw-r-- 1 d d 4417802560 Jan 3 19:10 model.safetensors
-rw-rw-r-- 1 d d 2479595168 Jan 3 19:10 unsloth.BF16.gguf <<<<< WORKS FINE.
-rw-rw-r-- 1 d d 255954592 Jan 3 19:10 unsloth.Q4_K_M.gguf <<<< DOES NOT WORK -INVALID MAGIC NUMBER