Upload folder using huggingface_hub (#3)

- 5a9c1b2a9a43e2cf12fc0c12abbfac6695bfda7b5183d5abc61390d80036790e (88d91b0d830871931d727068a35ad3b3be144c3b)
- 1a3cf829432d35b5ab5ff7e326a8db49db325db7a6451985f8806332e6a37529 (fa4ab1a44fc2fcdc84025de924b3942e1134d967)
- 98ba3df194d2ac1bb413b28d0618a6fddc17cc2377d1793229b8a4dc005864e4 (dbf24e7e3458854369ff9e0986339fe26a804ca8)

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/GritLM-GritLM-7B-bnb-8bit-smashed",
-                                                 trust_remote_code=True)
+                                                 trust_remote_code=True, device_map='auto')
     tokenizer = AutoTokenizer.from_pretrained("GritLM/GritLM-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/tmpkyzzzm49",
+  "_name_or_path": "/tmp/tmp7ro6rtjc",
   "architectures": [
     "MistralForCausalLM"
   ],
@@ -29,7 +29,7 @@
   "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/modelskz0k30m8",
+        "cache_dir": "/ceph/hdd/staff/charpent/.cache/models3rsn8hei",
         "batch_size": 1,
         "model_name": "GritLM/GritLM-7B",
         "pruning_ratio": 0.0,