Upload folder using huggingface_hub (#2)

- 689dea95499d4c78771a25cc1fef562b55a19119661cd175e121693d1e3237c1 (7b7a3ce9ed0b41cc21665b4b871bd4625618e160)
- ba3d0dc8510dfee6aaea024eb711edf9e3e9bc879fa1d8eb074a43d24c94480e (62ca8935564e37d92b43249e480d4112043cd087)

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/Intel-neural-chat-7b-v3-3-bnb-4bit-smashed",
- trust_remote_code=True)
+ trust_remote_code=True, device_map='auto')
  tokenizer = AutoTokenizer.from_pretrained("Intel/neural-chat-7b-v3-3")
  
  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/tmp_frfbyvc",
+ "_name_or_path": "/tmp/tmpgoz4271p",
  "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 sha256:51d5a0d2b72fa41de24dd10bf4996c13ad49037d38605fd0cad1e2e1146c0f9a
-size 4125687605
+oid sha256:d3010e255cdecaa669f74b9ffdc79ee225853703805bb82ff304c764d2c43b3f
+size 4450837496

smash_config.json

@@ -8,7 +8,7 @@
  "compilers": "None",
  "task": "text_text_generation",
  "device": "cuda",
- "cache_dir": "/ceph/hdd/staff/charpent/.cache/modelsvo295p_7",
+ "cache_dir": "/ceph/hdd/staff/charpent/.cache/modelse_62a_dp",
  "batch_size": 1,
  "model_name": "Intel/neural-chat-7b-v3-3",
  "pruning_ratio": 0.0,