Fixing the download strategy for ibm-fms (#1917)

# What does this PR do?

<!--
Congratulations! You've made it this far! You're not quite done yet
though.

Once merged, your PR is going to appear in the release notes with the
title you set, so make sure it's a great title that fully reflects the
extent of your awesome contribution.

Then, please replace this with a description of the change and which
issue is fixed (if applicable). Please also include relevant motivation
and context. List any dependencies (if any) that are required for this
change.

Once you're done, someone will review your PR shortly (see the section
"Who can review?" below to tag some potential reviewers). They may
suggest changes to make the code even better. If no one reviewed your PR
after a week has passed, don't hesitate to post a new comment
@-mentioning the same persons---sometimes notifications get lost.
-->

<!-- Remove if not applicable -->

Fixes # (issue)


## Before submitting
- [ ] This PR fixes a typo or improves the docs (you can dismiss the
other checks if that's the case).
- [ ] Did you read the [contributor
guideline](https://github.com/huggingface/transformers/blob/main/CONTRIBUTING.md#start-contributing-pull-requests),
      Pull Request section?
- [ ] Was this discussed/approved via a Github issue or the
[forum](https://discuss.huggingface.co/)? Please add a link
      to it if that's the case.
- [ ] Did you make sure to update the documentation with your changes?
Here are the
[documentation
guidelines](https://github.com/huggingface/transformers/tree/main/docs),
and
[here are tips on formatting
docstrings](https://github.com/huggingface/transformers/tree/main/docs#writing-source-documentation).
- [ ] Did you write any new necessary tests?


## Who can review?

Anyone in the community is free to review the PR once the tests have
passed. Feel free to tag
members/contributors who may be interested in your PR.

<!-- Your PR will be replied to more quickly if you can figure out the
right person to tag with @


@OlivierDehaene OR @Narsil

 -->

server/text_generation_server/cli.py

@@ -154,31 +154,27 @@ def download_weights(
         try:
             import json
 
-            medusa_head = hf_hub_download(
+            config = hf_hub_download(
-                model_id, revision=revision, filename="medusa_lm_head.safetensors"
-            )
-            medusa_config = hf_hub_download(
                 model_id, revision=revision, filename="config.json"
             )
-            with open(medusa_config, "r") as f:
+            with open(config, "r") as f:
                 config = json.load(f)
 
-            model_id = config["base_model_name_or_path"]
+            base_model_id = config.get("base_model_name_or_path", None)
-            revision = "main"
+            if base_model_id and base_model_id != model_id:
-            try:
+                try:
-                utils.weight_files(model_id, revision, extension)
+                    logger.info(f"Downloading parent model {base_model_id}")
-                logger.info(
+                    download_weights(
-                    f"Files for parent {model_id} are already present on the host. "
+                        model_id=base_model_id,
-                    "Skipping download."
+                        revision="main",
-                )
+                        extension=extension,
-                return
+                        auto_convert=auto_convert,
-            # Local files not found
+                        logger_level=logger_level,
-            except (
+                        json_output=json_output,
-                utils.LocalEntryNotFoundError,
+                        trust_remote_code=trust_remote_code,
-                FileNotFoundError,
+                    )
-                utils.EntryNotFoundError,
+                except Exception:
-            ):
+                    pass
-                pass
         except (utils.LocalEntryNotFoundError, utils.EntryNotFoundError):
             pass
 
@@ -195,31 +191,6 @@ def download_weights(
             if not extension == ".safetensors" or not auto_convert:
                 raise e
 
-    elif (Path(model_id) / "medusa_lm_head.safetensors").exists():
-        # Try to load as a local Medusa model
-        try:
-            import json
-
-            medusa_head = Path(model_id) / "medusa_lm_head.safetensors"
-            medusa_config = Path(model_id) / "config.json"
-            with open(medusa_config, "r") as f:
-                config = json.load(f)
-
-            model_id = config["base_model_name_or_path"]
-            revision = "main"
-            try:
-                utils.weight_files(model_id, revision, extension)
-                logger.info(
-                    f"Files for parent {model_id} are already present on the host. "
-                    "Skipping download."
-                )
-                return
-            # Local files not found
-            except (utils.LocalEntryNotFoundError, utils.EntryNotFoundError):
-                pass
-        except (utils.LocalEntryNotFoundError, utils.EntryNotFoundError):
-            pass
-
     elif (Path(model_id) / "adapter_config.json").exists():
         # Try to load as a local PEFT model
         try:
@@ -230,14 +201,43 @@ def download_weights(
             return
         except (utils.LocalEntryNotFoundError, utils.EntryNotFoundError):
             pass
+    elif (Path(model_id) / "config.json").exists():
+        # Try to load as a local Medusa model
+        try:
+            import json
+
+            config = Path(model_id) / "config.json"
+            with open(config, "r") as f:
+                config = json.load(f)
+
+            base_model_id = config.get("base_model_name_or_path", None)
+            if base_model_id:
+                try:
+                    logger.info(f"Downloading parent model {base_model_id}")
+                    download_weights(
+                        model_id=base_model_id,
+                        revision="main",
+                        extension=extension,
+                        auto_convert=auto_convert,
+                        logger_level=logger_level,
+                        json_output=json_output,
+                        trust_remote_code=trust_remote_code,
+                    )
+                except Exception:
+                    pass
+        except (utils.LocalEntryNotFoundError, utils.EntryNotFoundError):
+            pass
 
     # Try to see if there are local pytorch weights
     try:
         # Get weights for a local model, a hub cached model and inside the WEIGHTS_CACHE_OVERRIDE
-        local_pt_files = utils.weight_files(model_id, revision, ".bin")
+        try:
+            local_pt_files = utils.weight_files(model_id, revision, ".bin")
+        except Exception:
+            local_pt_files = utils.weight_files(model_id, revision, ".pt")
 
     # No local pytorch weights
-    except utils.LocalEntryNotFoundError:
+    except (utils.LocalEntryNotFoundError, utils.EntryNotFoundError):
         if extension == ".safetensors":
             logger.warning(
                 f"No safetensors weights found for model {model_id} at revision {revision}. "

server/text_generation_server/utils/hub.py

@@ -40,7 +40,6 @@ def _weight_hub_files_from_model_info(
         and "arguments" not in s.rfilename
         and "args" not in s.rfilename
         and "training" not in s.rfilename
-        and "medusa_lm_head" not in s.rfilename
     ]
 
 
@@ -57,7 +56,6 @@ def _weight_files_from_dir(d: Path, extension: str) -> List[str]:
         and "args" not in f
         and "adapter" not in f
         and "training" not in f
-        and "medusa_lm_head" not in f
     ]
     return filenames