Upload folder using huggingface_hub (#2)

- 667f26a97442e0cab35fc6d74c200867e5592826d08179fe5faa37a4a8368bb5 (5121b09796296fae9086c2aec1ed1450ddd212c5)
- 9813a9bdedc94d2e45faa1b8068c85046af8d32d865714ae65c9c9d838149127 (fbc49e745430c23c9f1ea3ee6733b42f2216b47f)

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/PygmalionAI-pygmalion-6b-bnb-8bit-smashed",
-                                                 trust_remote_code=True)
+                                                 trust_remote_code=True, device_map='auto')
     tokenizer = AutoTokenizer.from_pretrained("PygmalionAI/pygmalion-6b")
     
     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/tmpsvmut29m",
+  "_name_or_path": "/tmp/tmpxqtt9aa3",
   "activation_function": "gelu_new",
   "architectures": [
     "GPTJForCausalLM"
@@ -20,7 +20,7 @@
   "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/modelsgpv2t6fe",
+        "cache_dir": "/ceph/hdd/staff/charpent/.cache/models1gq3ru10",
         "batch_size": 1,
         "model_name": "PygmalionAI/pygmalion-6b",
         "pruning_ratio": 0.0,