add files

.gitignore

@@ -0,0 +1,5 @@
+llama.cpp/
+pytorch_model*.bin
+*.sha
+*.tar.gz
+tokenizer.model

Makefile

@@ -0,0 +1,47 @@
+MODEL_NAME=	open-llama-7b-v2
+PYTHON?=	python
+LLAMA_TAG=	db4047a
+LLAMA_TAR=	master-$(LLAMA_TAG).tar.gz
+HF_REPO=	openlm-research/open_llama_7b_v2
+HF_REF=		main
+HF_FILES=	pytorch_model-00001-of-00002.bin \
+		pytorch_model-00002-of-00002.bin \
+		tokenizer.model
+$(HF_FILES): 	SITE=	https://huggingface.co/$(HF_REPO)/resolve/$(HF_REF)
+$(LLAMA_TAR): 	SITE=	https://github.com/ggerganov/llama.cpp/archive/refs/tags
+
+FILES= $(HF_FILES) $(LLAMA_TAR)
+
+QUANTS=	f16 q4_0 q4_1 q5_0 q5_1 q8_0 \
+	q2_K \
+	q3_K_S q3_K_M q3_K_L \
+	q4_K_S q4_K_M \
+	q5_K_S q5_K_M \
+	q6_K
+
+MODEL_FILES=	$(addsuffix .bin,$(addprefix $(MODEL_NAME)-,$(QUANTS)))
+
+.PHONY: all
+all: $(MODEL_FILES) SHA256SUMS
+
+$(FILES):
+	curl -L -o $@ --url $(SITE)/$@
+
+llama.cpp: $(LLAMA_TAR)
+	mkdir -p $@
+	tar -xf $< --strip-components=1 -C $@
+
+llama.cpp/quantize: llama.cpp
+	$(MAKE) -C llama.cpp quantize
+
+$(MODEL_NAME)-f16.bin: $(HF_FILES) | llama.cpp
+	$(PYTHON) llama.cpp/convert.py --outtype f16 --outfile $@ .
+
+$(MODEL_NAME)-q%.bin: $(MODEL_NAME)-f16.bin | llama.cpp/quantize
+	llama.cpp/quantize $< $@ q$*
+
+%.sha: %
+	sha256sum $< > $@
+
+SHA256SUMS: $(addsuffix .sha,$(MODEL_FILES))
+	cat $^ > $@

README.md

@@ -1,3 +1,17 @@
 ---
 license: apache-2.0
 ---
+
+# ggml versions of OpenLLaMa 7B v2
+
+For use with [llama.cpp](https://github.com/ggerganov/llama.cpp).
+
+- Version: [Final version](https://github.com/openlm-research/open_llama#update-07072023)
+- Project: [OpenLLaMA: An Open Reproduction of LLaMA](https://github.com/openlm-research/open_llama)
+- Model: [openlm-research/open_llama_7b_v2](https://huggingface.co/openlm-research/open_llama_7b_v2)
+- llama.cpp 4,5,8-bit quantization: build 567(2d5db48) or later
+- llama.cpp newer quantization formats: build 616(99009e7) or later
+
+## Perplexity
+
+Coming soon...

SHA256SUMS

@@ -0,0 +1,15 @@
+e8638da6ae24ead336d7cf11ca822cbc2758ee6eba696259d7f44aacacdae28e  open-llama-7b-v2-f16.bin
+d70bb8fe5eefd194f55cc8dfd9277b57a20d611252d61c7aa2bee18cc9aae44f  open-llama-7b-v2-q4_0.bin
+a7bd7f6d19818ef85f81999965e9706974cfc5c3d2196b407fc1053bc1f0f71a  open-llama-7b-v2-q4_1.bin
+f753d0c9a8d28f4643452d5cb882376f62be295ab88fe41904373c77ec631b0a  open-llama-7b-v2-q5_0.bin
+09cb091143a3cdc70bfa956b49708f3e2b23d286a376b20cce90425f9ea33075  open-llama-7b-v2-q5_1.bin
+f2e016738892ac755a6bde4a8df435cc574edc26f4023bc6c5d9f865bae30624  open-llama-7b-v2-q8_0.bin
+af3221bab03e6f8c321d28ac473863b12d6731f29c953f2e529681bdff259718  open-llama-7b-v2-q2_K.bin
+f6039a48a5763d03c34d7f9f3516e49d44d55364a602944e365d924d280c28c7  open-llama-7b-v2-q3_K_S.bin
+8dca03aaf8aabaf2e10e6bd02ea33f3909e863d0f0e25e78af7a1ca23b743778  open-llama-7b-v2-q3_K_M.bin
+b74ce0b75e71a782a947309839cf8a4b989eb9572a352cd9c0c45b5cd62e0a1a  open-llama-7b-v2-q3_K_L.bin
+5de4cd3b6c44bd667324125b5163beed6a9b264360104e6bb0fb39c2538512a8  open-llama-7b-v2-q4_K_S.bin
+47903d1ce81eca3cd5f985ae5cae82769421d8a687287c8a8471abf9d186214f  open-llama-7b-v2-q4_K_M.bin
+38a4148fad35971d41df5a63d34fb8b538c0f5613736a3a732e36e0a4c1c3cf9  open-llama-7b-v2-q5_K_S.bin
+993894f8eb588882e1d63b2d1bd285b9d788cd297f2a491dfd62e482da354c08  open-llama-7b-v2-q5_K_M.bin
+a0be750e3702142b62629e1c7e79c625b1cbf76de1ded8bac1b8bba2e4a23826  open-llama-7b-v2-q6_K.bin

open-llama-7b-v2-f16.bin

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:e8638da6ae24ead336d7cf11ca822cbc2758ee6eba696259d7f44aacacdae28e
+size 13477822624

open-llama-7b-v2-q2_K.bin

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:af3221bab03e6f8c321d28ac473863b12d6731f29c953f2e529681bdff259718
+size 2866815136

open-llama-7b-v2-q3_K_L.bin

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:b74ce0b75e71a782a947309839cf8a4b989eb9572a352cd9c0c45b5cd62e0a1a
+size 3596828832

open-llama-7b-v2-q3_K_M.bin

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:8dca03aaf8aabaf2e10e6bd02ea33f3909e863d0f0e25e78af7a1ca23b743778
+size 3282256032

open-llama-7b-v2-q3_K_S.bin

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:f6039a48a5763d03c34d7f9f3516e49d44d55364a602944e365d924d280c28c7
+size 2948022432

open-llama-7b-v2-q4_0.bin

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:d70bb8fe5eefd194f55cc8dfd9277b57a20d611252d61c7aa2bee18cc9aae44f
+size 3825524896

open-llama-7b-v2-q4_1.bin

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:a7bd7f6d19818ef85f81999965e9706974cfc5c3d2196b407fc1053bc1f0f71a
+size 4238467232

open-llama-7b-v2-q4_K_M.bin

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:47903d1ce81eca3cd5f985ae5cae82769421d8a687287c8a8471abf9d186214f
+size 4080722080

open-llama-7b-v2-q4_K_S.bin

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:5de4cd3b6c44bd667324125b5163beed6a9b264360104e6bb0fb39c2538512a8
+size 3825524896

open-llama-7b-v2-q5_0.bin

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:f753d0c9a8d28f4643452d5cb882376f62be295ab88fe41904373c77ec631b0a
+size 4651409568

open-llama-7b-v2-q5_1.bin

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:09cb091143a3cdc70bfa956b49708f3e2b23d286a376b20cce90425f9ea33075
+size 5064351904

open-llama-7b-v2-q5_K_M.bin

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:993894f8eb588882e1d63b2d1bd285b9d788cd297f2a491dfd62e482da354c08
+size 4782874784

open-llama-7b-v2-q5_K_S.bin

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:38a4148fad35971d41df5a63d34fb8b538c0f5613736a3a732e36e0a4c1c3cf9
+size 4651409568

open-llama-7b-v2-q6_K.bin

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:a0be750e3702142b62629e1c7e79c625b1cbf76de1ded8bac1b8bba2e4a23826
+size 5528912032

open-llama-7b-v2-q8_0.bin

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:f2e016738892ac755a6bde4a8df435cc574edc26f4023bc6c5d9f865bae30624
+size 7129063584