Upload FalconForCausalLM

config.json

@@ -0,0 +1,55 @@
+{
+  "_name_or_path": "tiiuae/falcon-40b",
+  "activation": "gelu",
+  "alibi": false,
+  "apply_residual_connection_post_layernorm": false,
+  "architectures": [
+    "FalconForCausalLM"
+  ],
+  "attention_dropout": 0.0,
+  "auto_map": {
+    "AutoConfig": "tiiuae/falcon-40b--configuration_falcon.FalconConfig",
+    "AutoModel": "tiiuae/falcon-40b--modeling_falcon.FalconModel",
+    "AutoModelForCausalLM": "tiiuae/falcon-40b--modeling_falcon.FalconForCausalLM",
+    "AutoModelForQuestionAnswering": "tiiuae/falcon-40b--modeling_falcon.FalconForQuestionAnswering",
+    "AutoModelForSequenceClassification": "tiiuae/falcon-40b--modeling_falcon.FalconForSequenceClassification",
+    "AutoModelForTokenClassification": "tiiuae/falcon-40b--modeling_falcon.FalconForTokenClassification"
+  },
+  "bias": false,
+  "bos_token_id": 11,
+  "eos_token_id": 11,
+  "ffn_hidden_size": 32768,
+  "hidden_dropout": 0.0,
+  "hidden_size": 8192,
+  "initializer_range": 0.02,
+  "layer_norm_epsilon": 1e-05,
+  "max_position_embeddings": 2048,
+  "model_type": "falcon",
+  "multi_query": true,
+  "new_decoder_architecture": true,
+  "num_attention_heads": 128,
+  "num_hidden_layers": 60,
+  "num_kv_heads": 8,
+  "parallel_attn": true,
+  "quantization_config": {
+    "_load_in_4bit": true,
+    "_load_in_8bit": false,
+    "bnb_4bit_compute_dtype": "bfloat16",
+    "bnb_4bit_quant_storage": "uint8",
+    "bnb_4bit_quant_type": "nf4",
+    "bnb_4bit_use_double_quant": true,
+    "llm_int8_enable_fp32_cpu_offload": false,
+    "llm_int8_has_fp16_weight": false,
+    "llm_int8_skip_modules": null,
+    "llm_int8_threshold": 6.0,
+    "load_in_4bit": true,
+    "load_in_8bit": false,
+    "quant_method": "bitsandbytes"
+  },
+  "rope_scaling": null,
+  "rope_theta": 10000.0,
+  "torch_dtype": "bfloat16",
+  "transformers_version": "4.40.2",
+  "use_cache": true,
+  "vocab_size": 65024
+}

generation_config.json

@@ -0,0 +1,6 @@
+{
+  "_from_model_config": true,
+  "bos_token_id": 11,
+  "eos_token_id": 11,
+  "transformers_version": "4.40.2"
+}

model-00001-of-00005.safetensors

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:c398c0511b4bb573cf822fb8392db89bac23447ab548548caf04e89a85966f0e
+size 4995472631

model-00002-of-00005.safetensors

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:34fbef1c4450b3b6354e418aab60caf9af45716b1695e26558be057a21f6587a
+size 4908335392

model-00003-of-00005.safetensors

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:fa7020657c25df2fc9de076850602a96541cd15ff3ea40a5767f36df6be9cbbd
+size 4908335397

model-00004-of-00005.safetensors

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:a9ac10955cf6e03eda7625a6703f0b97cda1607469cb039ca6b5db7a9f5c254d
+size 4908335392

model-00005-of-00005.safetensors

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:54e0b42ec9f8ebb221530ef96ec19a1535675ebc9d2149aa05217abb74bb991d
+size 2380630248