Upload folder using huggingface_hub (#3)

- 53c5b2ec1b422daf544a0cd8c45c7e777ed646146c2787f445303436541fca1a (cbffb06d3c5974eb0d5778c3cb2945b42abc4a9a)
- c7b657bafcb5929e66fc62d2a56e7929cc838b33ac4d09d20de4684a38096019 (8c7d62d9b665cacd39a1cae351e8350ef0b8cfc9)
- b528ec290745a2609a64c92b0411895033a3cee271b45deea33ee05e84475c6e (87969cfaa17c4538e643990f2f1b9391a1ada36e)

README.md

@@ -34,7 +34,7 @@ tags:
 
 ## Results
 
-Detailed efficiency metrics coming soon!
+![image info](./plots.png)
 
 **Frequently Asked Questions**
 - ***How does the compression work?*** The model is compressed with llm-int8.
@@ -61,7 +61,7 @@ You can run the smashed model with these steps:
    from transformers import AutoModelForCausalLM, AutoTokenizer
 
     model = AutoModelForCausalLM.from_pretrained("PrunaAI/BioMistral-BioMistral-7B-bnb-8bit-smashed",
-                                                 trust_remote_code=True)
+                                                 trust_remote_code=True, device_map='auto')
     tokenizer = AutoTokenizer.from_pretrained("BioMistral/BioMistral-7B")
     
     input_ids = tokenizer("What is the color of prunes?,", return_tensors='pt').to(model.device)["input_ids"]

config.json

@@ -1,5 +1,5 @@
 {
-  "_name_or_path": "/tmp/tmpe9r_max1",
+  "_name_or_path": "/tmp/tmpodq0q40z",
   "architectures": [
     "MistralForCausalLM"
   ],
@@ -18,7 +18,7 @@
   "quantization_config": {
     "bnb_4bit_compute_dtype": "bfloat16",
     "bnb_4bit_quant_type": "fp4",
-    "bnb_4bit_use_double_quant": true,
+    "bnb_4bit_use_double_quant": false,
     "llm_int8_enable_fp32_cpu_offload": false,
     "llm_int8_has_fp16_weight": false,
     "llm_int8_skip_modules": [

smash_config.json

@@ -8,7 +8,7 @@
         "compilers": "None",
         "task": "text_text_generation",
         "device": "cuda",
-        "cache_dir": "/ceph/hdd/staff/charpent/.cache/modelsm8ayynki",
+        "cache_dir": "/ceph/hdd/staff/charpent/.cache/modelsnghe_ab5",
         "batch_size": 1,
         "model_name": "BioMistral/BioMistral-7B",
         "pruning_ratio": 0.0,