mradermacher

/

model_requests

like 159

It unfortinately failed with the following error:

Terminus-LFM2.5-350m    WARNING:gguf.gguf_writer:Duplicated key name 'lfm2.attention.layer_norm_rms_epsilon', overwriting it with new value 1e-05 of type FLOAT32
Terminus-LFM2.5-350m    Traceback (most recent call last):
Terminus-LFM2.5-350m      File "/llmjob/llama.cpp/convert_hf_to_gguf.py", line 13127, in <module>
Terminus-LFM2.5-350m        main()
Terminus-LFM2.5-350m      File "/llmjob/llama.cpp/convert_hf_to_gguf.py", line 13121, in main
Terminus-LFM2.5-350m        model_instance.write()
Terminus-LFM2.5-350m      File "/llmjob/llama.cpp/convert_hf_to_gguf.py", line 945, in write
Terminus-LFM2.5-350m        self.prepare_metadata(vocab_only=False)
Terminus-LFM2.5-350m      File "/llmjob/llama.cpp/convert_hf_to_gguf.py", line 1064, in prepare_metadata
Terminus-LFM2.5-350m        super().prepare_metadata(vocab_only=vocab_only)
Terminus-LFM2.5-350m      File "/llmjob/llama.cpp/convert_hf_to_gguf.py", line 935, in prepare_metadata
Terminus-LFM2.5-350m        self.set_gguf_parameters()
Terminus-LFM2.5-350m      File "/llmjob/llama.cpp/convert_hf_to_gguf.py", line 11858, in set_gguf_parameters
Terminus-LFM2.5-350m        self._add_feed_forward_length()
Terminus-LFM2.5-350m      File "/llmjob/llama.cpp/convert_hf_to_gguf.py", line 11831, in _add_feed_forward_length
Terminus-LFM2.5-350m        ff_dim = self.hparams["block_ff_dim"]
Terminus-LFM2.5-350m    KeyError: 'block_ff_dim'
Terminus-LFM2.5-350m    job finished, status 1
Terminus-LFM2.5-350m    job-done<0 Terminus-LFM2.5-350m noquant 1>
Terminus-LFM2.5-350m
Terminus-LFM2.5-350m    NAME: Terminus-LFM2.5-350m
Terminus-LFM2.5-350m    TIME: Sun Apr 12 06:57:12 2026
Terminus-LFM2.5-350m    WORKER: leia

I think u need to update llama.cpp for the new LFM2.5-350m arch