Upload folder using huggingface_hub (#2)

- 18161e7b08edc4085b9ef0f118f85f47a4f94cef452279d4da41470b85a864f9 (a2529296d97384c6d38716f6144838a120bdd565)
- 589fe748081eca9c943b63e2757ff8fd212cf87899addcd3e441cc706f8ecf8f (e079f7fbfc61c592d800f15299c7d6d39448e043)
- 58236eeaffb96a338ebb21077bd7f8be057e412bb40e91a81db4d692836dec7e (47a17d4a00ffaaea6f8cd94448fd45d86ab06d11)

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/norallm-normistral-7b-warm-bnb-8bit-smashed",
-                                                 trust_remote_code=True)
+                                                 trust_remote_code=True, device_map='auto')
     tokenizer = AutoTokenizer.from_pretrained("norallm/normistral-7b-warm")
     
     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/tmpf256uewp",
+  "_name_or_path": "/tmp/tmpg1qf4342",
   "architectures": [
     "MistralForCausalLM"
   ],
@@ -19,7 +19,7 @@
   "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/models5bqts955",
+        "cache_dir": "/ceph/hdd/staff/charpent/.cache/models6ifiukjd",
         "batch_size": 1,
         "model_name": "norallm/normistral-7b-warm",
         "pruning_ratio": 0.0,