Upload folder using huggingface_hub (#2)

- 0e8bd0e4c82fc963488a4a8c41c0fc9a969c7561ac57d7892c8eb141fd6f6a38 (5364dfe56364615878b55c3766e49158d111b614)
- eee47c1902385854367d635bb2fde626fceef2b0ecfe1f1365cdbe6c9cc8023b (02c44f503f74282d1b13a00bb8c0667bcd803cae)
- a32abceed8dda1b725f2393d42805dccaedb35e39681c814cf16b7f6ea4ffec7 (7ca6652fb1f161c3382f0a301ed53e67cd77f266)

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/openchat-openchat_3.5-bnb-8bit-smashed",
-                                                 trust_remote_code=True)
+                                                 trust_remote_code=True, device_map='auto')
     tokenizer = AutoTokenizer.from_pretrained("openchat/openchat_3.5")
     
     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/tmpwl3n292i",
+  "_name_or_path": "/tmp/tmpe0_vj5t5",
   "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/models5v8zez5v",
+        "cache_dir": "/ceph/hdd/staff/charpent/.cache/modelsk6d8bji0",
         "batch_size": 1,
         "model_name": "openchat/openchat_3.5",
         "pruning_ratio": 0.0,