Upload 2 files

model_tools.md

@@ -26,6 +26,9 @@ Tools to enhance LLM quantizations and merging
 # [fp32_to_fp16.py](https://huggingface.co/spaces/Naphula/model_tools/blob/main/fp32_to_fp16.py)
 - Converts FP32 to FP16 safetensors
 
+# [pytorch_to_safetensors.py](https://huggingface.co/spaces/Naphula/model_tools/blob/main/pytorch_to_safetensors.py)
+- Converts pytorch bin to safetensors format
+
 # [textonly_ripper_v2.py](https://huggingface.co/spaces/Naphula/model_tools/blob/main/textonly_ripper_v2.py)
 - Converts a sharded, multimodal (text and vision) model into a text-only version. Readme at [textonly_ripper.md](https://huggingface.co/spaces/Naphula/model_tools/blob/main/textonly_ripper.md)
 

pytorch_to_safetensors.py

@@ -0,0 +1,49 @@
+import os
+import torch
+
+# --- AGGRESSIVE FIX: Bypass Security Check ---
+# We must import these modules specifically to patch the function where it is used
+import transformers.modeling_utils
+import transformers.utils.import_utils
+
+# Disable the check in both locations
+transformers.modeling_utils.check_torch_load_is_safe = lambda: None
+transformers.utils.import_utils.check_torch_load_is_safe = lambda: None
+# ---------------------------------------------
+
+from transformers import AutoModelForCausalLM, AutoTokenizer
+
+# 1. Path to your local PyTorch model
+input_path = r"B:\7B\!models--Gryphe--Tiamat-7b" 
+
+# 2. Path where you want the SafeTensors version
+output_path = r"B:\7B\!models--Gryphe--Tiamat-7b\safe"
+
+print(f"Loading model from {input_path}...")
+
+# Load the model
+model = AutoModelForCausalLM.from_pretrained(
+    input_path,
+    torch_dtype=torch.bfloat16,
+    device_map="cpu",
+    low_cpu_mem_usage=True
+)
+
+# Load the tokenizer
+tokenizer = AutoTokenizer.from_pretrained(input_path)
+
+print(f"Saving to {output_path}...")
+
+if not os.path.exists(output_path):
+    os.makedirs(output_path)
+
+# 3. Save with safe_serialization=True
+model.save_pretrained(
+    output_path, 
+    safe_serialization=True, 
+    max_shard_size="5GB"
+)
+
+tokenizer.save_pretrained(output_path)
+
+print("Conversion complete.")