nikola/llama_cpp
1
0
Fork 0
Code Issues Pull Requests Actions 4 Packages Projects Releases Wiki Activity

model: Granite Speech Plus (#24818)

Browse Source 
* feat: Add conversion support for Granite Speech Plus

Branch: GraniteSpeechPlus
AI-usage: full (Bob, OpenCode + Qwen3.6-35b)
Signed-off-by: Gabe Goodhart <ghart@us.ibm.com>

* feat: Extend granite_speech to support plus multi-layer concatenation

Branch: GraniteSpeechPlus
AI-usage: draft (Bob, OpenCode + Qwen3.6-35b)
Signed-off-by: Gabe Goodhart <ghart@us.ibm.com>

* fix(conversion): Fix plural naming for feature_layers for audio

Branch: GraniteSpeechPlus
AI-usage: none
Signed-off-by: Gabe Goodhart <ghart@us.ibm.com>

* fix(mtmd): Align feature_layer usage and naming everywhere

Branch: GraniteSpeechPlus
AI-usage: none
Signed-off-by: Gabe Goodhart <ghart@us.ibm.com>

* style: Use fstring for log

Signed-off-by: Gabe Goodhart <ghart@us.ibm.com>

Co-authored-by: Xuan-Son Nguyen <thichthat@gmail.com>

---------

Co-authored-by: Xuan-Son Nguyen <thichthat@gmail.com>
This commit is contained in:
Gabe Goodhart
2026-06-23 04:03:31 -06:00
committed by GitHub
parent 7c908502ea
commit a3900a6694
10 changed files with 87 additions and 20 deletions
Download Patch File Download Diff File Expand all files Collapse all files

2
conversion/__init__.py
View File

@@ -96,6 +96,7 @@ TEXT_MODEL_MAP: dict[str, str] = {
"GraniteMoeHybridForCausalLM": "granite",
"GraniteMoeSharedForCausalLM": "granite",
"GraniteSpeechForConditionalGeneration": "granite",
"GraniteSpeechPlusForConditionalGeneration": "granite",
"Grok1ForCausalLM": "grok",
"GrokForCausalLM": "grok",
"GroveMoeForCausalLM": "grovemoe",
@@ -261,6 +262,7 @@ MMPROJ_MODEL_MAP: dict[str, str] = {
"GlmasrModel": "ultravox",
"Granite4VisionForConditionalGeneration": "granite",
"GraniteSpeechForConditionalGeneration": "granite",
"GraniteSpeechPlusForConditionalGeneration": "granite",
"HunYuanVLForConditionalGeneration": "hunyuan",
"Idefics3ForConditionalGeneration": "smolvlm",
"InternVisionModel": "internvl",

28
conversion/granite.py
View File

@@ -348,6 +348,34 @@ class GraniteSpeechMmprojModel(MmprojModel):
yield from super().modify_tensors(data_torch, name, bid)
@ModelBase.register("GraniteSpeechPlusForConditionalGeneration")
class GraniteSpeechPlusMmprojModel(GraniteSpeechMmprojModel):
"""Conversion for GraniteSpeechPlus - extends GraniteSpeech with feature layer concatenation"""
has_vision_encoder = False
has_audio_encoder = True
def set_gguf_parameters(self):
assert self.hparams_audio is not None
super().set_gguf_parameters()
# Add feature_layer if present in encoder config
if feature_layers := self.hparams_audio.get("cat_hidden_layers"):
self.gguf_writer.add_audio_feature_layers(feature_layers)
logger.info(f"gguf: audio feature_layers = {feature_layers}")
# Validate projector dimension matches concatenated encoder output
hidden_dim = self.hparams_audio["hidden_dim"]
expected_dim = hidden_dim * (len(feature_layers) + 1)
projector_dim = self.global_config["projector_config"]["encoder_hidden_size"]
if projector_dim != expected_dim:
raise ValueError(
f"Projector encoder_hidden_size ({projector_dim}) does not match "
f"expected concatenated dimension ({expected_dim}). "
f"Expected: hidden_dim ({hidden_dim}) * (len(feature_layers) + 1) = {expected_dim}"
)
@ModelBase.register("Granite4VisionForConditionalGeneration")
class Granite4VisionMmprojModel(MmprojModel):
has_vision_encoder = True