Upload folder using huggingface_hub (#2)

- 55387347a76163a5be6cda508d828f53f06649119cde2f4753a214589218e153 (83caae8b1eec8aaaa30eef9a1626daf64cf3eba4)
- 9c194b44fef7ee7172ade634969292a5d96a3a3f36627dc554565338c47d871f (9fab2b712ec21266a67961758e2ce59b0a5208a1)

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/maywell-Synatra-7B-v0.3-dpo-bnb-4bit-smashed",
-                                                 trust_remote_code=True)
+                                                 trust_remote_code=True, device_map='auto')
     tokenizer = AutoTokenizer.from_pretrained("maywell/Synatra-7B-v0.3-dpo")
     
     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/tmp4ea3aiu8",
+  "_name_or_path": "/tmp/tmpkz3vybnr",
   "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": [

model.safetensors

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:e76eed93c86bef7c442129cf1a905fc0d8e7fb5b973095c2342f104901716392
-size 4125720391
+oid sha256:207bdbf8221bca326585a309f7a5fce166934ead9b2cae29500c44a74dc36ac8
+size 4450870264

smash_config.json

@@ -8,7 +8,7 @@
         "compilers": "None",
         "task": "text_text_generation",
         "device": "cuda",
-        "cache_dir": "/ceph/hdd/staff/charpent/.cache/models0bz4epz1",
+        "cache_dir": "/ceph/hdd/staff/charpent/.cache/modelsf5a4uuiq",
         "batch_size": 1,
         "model_name": "maywell/Synatra-7B-v0.3-dpo",
         "pruning_ratio": 0.0,