vbatts/llama.cpp
1
0
Fork 0
Code Issues Pull requests Projects Releases Packages Wiki Activity Actions

llama : add support for the cohere2 model architecture (#10900)

Browse source
This commit is contained in:
DAN™ 2025-01-04 09:33:31 -05:00 • committed by GitHub
parent 78c6785175
commit 46be942214
No known key found for this signature in database
GPG key ID: B5690EEEBB952194
6 changed files with 221 additions and 0 deletions
Download patch file Download diff file Expand all files Collapse all files

18
convert_hf_to_gguf.py
View file

@ -3373,6 +3373,24 @@ class CommandR2Model(Model):
self.gguf_writer.add_rope_scaling_type(gguf.RopeScalingType.NONE)
@Model.register("Cohere2ForCausalLM")
class Cohere2Model(Model):
model_arch = gguf.MODEL_ARCH.COHERE2
def set_gguf_parameters(self):
super().set_gguf_parameters()
self.gguf_writer.add_logit_scale(self.hparams["logit_scale"])
self.gguf_writer.add_sliding_window(self.hparams["sliding_window"])
self.gguf_writer.add_vocab_size(self.hparams["vocab_size"])
rotary_pct = self.hparams["rotary_pct"]
hidden_size = self.hparams["hidden_size"]
num_attention_heads = self.hparams["num_attention_heads"]
self.gguf_writer.add_rope_dimension_count(int(rotary_pct * (hidden_size // num_attention_heads)))
self.gguf_writer.add_rope_scaling_type(gguf.RopeScalingType.NONE)
@Model.register("OlmoForCausalLM")
@Model.register("OLMoForCausalLM")
class OlmoModel(Model):