Upload folder using huggingface_hub (#2)

- 4937ca114e9d0420436bff25d85866749f5f06ee9391545bf1d0250c753df93f (c77cc41207f7548f46c47f5d94ce630b5fc17ddc)
- 420f204adb9aea5a02e4772a8ae303c15537caac490bbfab0c4d51848c16dccb (11db7ffef9eb0b7794f48127174e11344ecdf29c)

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/kyujinpy-Ko-PlatYi-6B-bnb-4bit-smashed",
-                                                 trust_remote_code=True)
+                                                 trust_remote_code=True, device_map='auto')
     tokenizer = AutoTokenizer.from_pretrained("kyujinpy/Ko-PlatYi-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/tmpb4f5gal2",
+  "_name_or_path": "/tmp/tmpp_w3xu_v",
   "architectures": [
     "LlamaForCausalLM"
   ],
@@ -21,7 +21,7 @@
   "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": [

model.safetensors

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:b74eb5fe320f0d587b815b613287b66b38b9d89c1a58ab5f47025d9b8863c422
-size 4142636767
+oid sha256:5cc0b8155716198f3cf502416a73968d3fe9273450219ef091cb02e70b5434ed
+size 4400505856

smash_config.json

@@ -8,7 +8,7 @@
         "compilers": "None",
         "task": "text_text_generation",
         "device": "cuda",
-        "cache_dir": "/ceph/hdd/staff/charpent/.cache/modelste2ian_q",
+        "cache_dir": "/ceph/hdd/staff/charpent/.cache/modelsaygyx0ft",
         "batch_size": 1,
         "model_name": "kyujinpy/Ko-PlatYi-6B",
         "pruning_ratio": 0.0,