initial commit.

.gitignore

@@ -0,0 +1,7 @@
+convert.py
+llama.cpp*/
+pytorch_model.bin
+*.sha
+*.tar.gz
+tokenizer.model
+config.json

Makefile

@@ -0,0 +1,42 @@
+MODEL_NAME=	flan-open-llama-3b
+PYTHON?=	python
+LLAMA_TAG=	7487137
+LLAMA_TAR=	master-$(LLAMA_TAG).tar.gz
+LLAMA_DIR=	llama.cpp-master-$(LLAMA_TAG)
+LLAMA_FLAGS=	LLAMA_NO_K_QUANTS=1
+HF_REPO=	conceptofmind/Flan-Open-Llama-3b
+HF_REF=		main
+HF_FILES=	pytorch_model.bin \
+		tokenizer.model \
+		config.json
+$(HF_FILES): 	SITE=	https://huggingface.co/$(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
+MODEL_FILES=	$(foreach q,$(QUANTS),$(MODEL_NAME)-$(q).bin)
+
+.PHONY: all
+all: $(MODEL_FILES) SHA256SUMS
+
+$(FILES):
+	curl -L -o $@ --url $(SITE)/$@
+
+$(LLAMA_DIR): | $(LLAMA_TAR)
+	tar -xf $(LLAMA_TAR)
+
+$(LLAMA_DIR)/quantize: | $(LLAMA_DIR)
+	$(MAKE) -C $(LLAMA_DIR) $(LLAMA_FLAGS) quantize
+
+$(MODEL_NAME)-f16.bin: $(HF_FILES) | $(LLAMA_DIR)
+	$(PYTHON) $(LLAMA_DIR)/convert.py --outtype f16 --outfile $@ .
+
+$(MODEL_NAME)-q%.bin: $(MODEL_NAME)-f16.bin $(LLAMA_DIR)/quantize
+	$(LLAMA_DIR)/quantize $< $@ q$*
+
+%.sha: %
+	sha256sum $< > $@
+
+SHA256SUMS: $(addsuffix .sha,$(MODEL_FILES))
+	cat $^ > $@

README.md

@@ -1,3 +1,23 @@
 ---
 license: apache-2.0
 ---
+
+# ggml versions of Flan-Open-Llama-3b
+
+- Announcement: [Tweet by @EnricoShippole](https://twitter.com/EnricoShippole/status/1661756166248996867) ("open-source")
+- Model: [conceptofmind/Flan-Open-Llama-3b](https://huggingface.co/conceptofmind/Flan-Open-Llama-3b)
+- Base Model: [openlm-research/open_llama_3b](https://huggingface.co/openlm-research/open_llama_3b) [OpenLLaMA: An Open Reproduction of LLaMA](https://github.com/openlm-research/open_llama) (Apache 2.0)
+- Dataset: [FLAN](https://github.com/google-research/FLAN) (Apache 2.0)
+- [llama.cpp](https://github.com/ggerganov/llama.cpp): build 607(ffb06a3) or later
+- Type: instruct
+
+## Use with llama.cpp
+
+Support is now merged to master branch.
+
+## K-quants
+
+There are now more quantization types in llama.cpp, some lower than 4 bits.
+Currently these are not well supported because of technical reasons.
+If you want to use them, you have to build llama.cpp (from build 829 (ff5d58f)) with the `LLAMA_QKK_64` Make or CMake variable enabled (see PR [#2001](https://github.com/ggerganov/llama.cpp/pull/2001)).
+Then you can quantize the F16 or maybe Q8_0 version to what you want.

SHA256SUMS

@@ -0,0 +1,6 @@
+b3d6654431552b1600a5337d49f18230b525297e6f5dfcebb1e58aa53742f952  flan-open-llama-3b-f16.bin
+78e32dd9f16e2ff835f3254bf6b5b51116e6c86c99a6ebaa2de18598704489fc  flan-open-llama-3b-q4_0.bin
+0aa777b9c0837ccb57aa7ed7a3f180b1b3cead44bfb0a8e7f02d7af84ab67459  flan-open-llama-3b-q4_1.bin
+6d76ccd0a016435984650e241a5168d271dace34333b5d87fbd5760e60ed34d4  flan-open-llama-3b-q5_0.bin
+808870854a43b9c3392f156f6975de30fd5189ce6ecdc244dabeb96777e85059  flan-open-llama-3b-q5_1.bin
+3b9cc39620e8b52343b29a41f58ddf34fd7dbd601d748134e839f8b595e6ed55  flan-open-llama-3b-q8_0.bin

flan-open-llama-3b-f16.bin

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:b3d6654431552b1600a5337d49f18230b525297e6f5dfcebb1e58aa53742f952
+size 6853758208

flan-open-llama-3b-q4_0.bin

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:78e32dd9f16e2ff835f3254bf6b5b51116e6c86c99a6ebaa2de18598704489fc
+size 1928446208

flan-open-llama-3b-q4_1.bin

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:0aa777b9c0837ccb57aa7ed7a3f180b1b3cead44bfb0a8e7f02d7af84ab67459
+size 2142590208

flan-open-llama-3b-q5_0.bin

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:6d76ccd0a016435984650e241a5168d271dace34333b5d87fbd5760e60ed34d4
+size 2356734208

flan-open-llama-3b-q5_1.bin

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:808870854a43b9c3392f156f6975de30fd5189ce6ecdc244dabeb96777e85059
+size 2570878208

flan-open-llama-3b-q8_0.bin

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:3b9cc39620e8b52343b29a41f58ddf34fd7dbd601d748134e839f8b595e6ed55
+size 3641598208