Skip to content

Fix issues during StableLM models conversion due to config.json changes #5703

New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

Sign up for GitHub

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Merged
merged 8 commits into from
Feb 25, 2024
Merged

Fix issues during StableLM models conversion due to config.json changes #5703

Changes from 6 commits
Commits
Show all changes
8 commits
Select commit Hold shift + click to select a range
8a52999
Fix issues during StableLM models conversion
aahouzi Feb 24, 2024
d4a952e
Fix hard coded layer_norm_eps
aahouzi Feb 24, 2024
a6dbec8
Support layer_norm_eps for LlavaStableLM
aahouzi Feb 24, 2024
79959ca
Add missing parenthesis
aahouzi Feb 24, 2024
1989186
Support rotary_factor for LlavaStableLM
aahouzi Feb 24, 2024
a0c57b3
fix typo
cebtenzzre Feb 24, 2024
2e0ae2e
Add StableLMEpochForCausalLM for safety
aahouzi Feb 25, 2024
371e955
Add StableLMEpochForCausalLM for safety 2
aahouzi Feb 25, 2024
File filter

Filter by extension

Filter by extension
Conversations
Failed to load comments.
Loading
Jump to
Jump to file
Failed to load files.
Loading
Diff view
Diff view
9 changes: 5 additions & 4 deletions convert-hf-to-gguf.py
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -192,7 +192,7 @@ def from_model_architecture(model_architecture):
return RefactModel
if model_architecture == "PersimmonForCausalLM":
return PersimmonModel
if model_architecture in ("StableLMEpochForCausalLM", "LlavaStableLMEpochForCausalLM"):
if model_architecture in ("StableLmForCausalLM", "LlavaStableLMEpochForCausalLM"):
return StableLMModel
if model_architecture == "QWenLMHeadModel":
return QwenModel
Expand Down Expand Up @@ -253,7 +253,7 @@ def _get_model_architecture(self) -> gguf.MODEL_ARCH:
return gguf.MODEL_ARCH.REFACT
if arch == "PersimmonForCausalLM":
return gguf.MODEL_ARCH.PERSIMMON
if arch in ("StableLMEpochForCausalLM", "LlavaStableLMEpochForCausalLM"):
if arch in ("StableLmForCausalLM", "LlavaStableLMEpochForCausalLM"):
return gguf.MODEL_ARCH.STABLELM
if arch == "QWenLMHeadModel":
return gguf.MODEL_ARCH.QWEN
Expand Down Expand Up @@ -1074,10 +1074,11 @@ def set_gguf_parameters(self):
self.gguf_writer.add_embedding_length(hparams["hidden_size"])
self.gguf_writer.add_block_count(block_count)
self.gguf_writer.add_feed_forward_length(hparams["intermediate_size"])
self.gguf_writer.add_rope_dimension_count(int(hparams["rope_pct"] * (hparams["hidden_size"] // hparams["num_attention_heads"])))
rotary_factor = self.find_hparam(["partial_rotary_factor", "rope_pct"])
self.gguf_writer.add_rope_dimension_count(int(rotary_factor * (hparams["hidden_size"] // hparams["num_attention_heads"])))
self.gguf_writer.add_head_count(hparams["num_attention_heads"])
self.gguf_writer.add_parallel_residual(hparams["use_parallel_residual"] if "use_parallel_residual" in hparams else True)
self.gguf_writer.add_layer_norm_eps(1e-5)
self.gguf_writer.add_layer_norm_eps(self.find_hparam(["layer_norm_eps", "norm_eps"]))


class MixtralModel(Model):
Expand Down